Find interesting articles on data science and creating value with analytics written by the Kausa team.

Welcome to Kausa Blog

João Sousa

Sales Manager

BI needs to evolve to create actionable insights and business impact

Die or thrive - what's next for dashboards

Dashboards have been around for over 20 years as the go-to product for business intelligence. Back then, businesses didn’t collect much data and dashboards enabled both reporting (i.e., metrics monitoring) and actionable insights (i.e., diagnostic analytics).

However, a lot has changed in the meantime. Businesses collect more data than ever. In fact, 90% of the world’s data has been collected over the last 2 years. Businesses have millions of data points across products, marketing, etc. Moreover, the needs and expectations have significantly increased. Business stakeholders need to be agile, hence expect quick and comprehensive answers in order to make decisions before it’s too late.

This new complexity and expectations make dashboards fall short for many use cases, especially around data exploration and diagnostic analytics. Many teams navigate around multiple dashboards as if they are navigating in a foggy forest hoping to have a light to help them uncover a  path.

_“Dashboards started out as reporting tools for measuring the health of the business. They’re great for this purpose. The problem arises when you extend this simple purpose into data exploration.”_

## Where dashboards are valuable

Some vendors claim that dashboards are dead, which I consider to be exaggerated. Dashboards are here to stay. They are very valuable in business monitoring, understanding how key business metrics are evolving, and to catch high-level trends. (e.g., marketing spending across regions). They enable people to observe what is happening in the business over time. 

## Where dashboards fall short

The problem raises as dashboards are trying to be repurposed. Every time business stakeholders ask data analysts for a new dashboard on a key business metric, what they actually say is “We need to understand what’s driving this key business metric and what can we do to improve it.” Adding more charts looking at the business metric from different angles is likely useless. All these dashboards won’t answer the most fundamental question: “Now what?”.

Oftentimes data people believe that a comprehensive, well-designed dashboard is a very capable data product that gives business stakeholders endless possibilities to answer any questions they might have.

The reality? Frequently, it’s far from it. Oftentimes business stakeholders and data analysts run into an insights dead end. To reach the “so what”, teams first need to understand what triggered the change in the metric.  And the process of diagnosing a key business metric change can be very tedious. And outcome looks something like this:

![dashboard-snockered.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/dashboard_snockered_77ae73006e.png)

The outcome of dashboard-based analysis (image by Anderson for eQuest)

## The new reality — needs and requirements

The new reality in the amount of data and business speed is making these shortcomings more evident and these 3 requirements critical:

- **Super-human speed** — business stakeholders expect quick answers to performance changes to be agile and stay ahead of the competition. Teams can’t wait days or weeks to get answers.
- **Comprehensiveness** —with more data points than ever, stakeholders expect more granular insights to drive better decisions. This requires leveraging all the available data to connect the dots and find granular insights.
- **Focus** — Remember the foggy forest and the light analogy? Nowadays BI platforms are expected to help teams fast-forward their analysis, by pinpointing where to look. 

## Consequences of dashboard-based workflows

Relying on reports and dashboards that only present what happened with key metrics and having to drill down to why these changes happened manually doesn’t meet the expectations anymore. Nowadays organizations set carefully designed metrics, expect to be able to monitor these on a daily basis and make fast decisions based on the latest numbers. 

Continuing with traditional approaches results in 4 main negative consequences, leading to significant lost business value:

- **Reactive** —   Most of the time business teams spot a sharp increase or a decrease on a dashboard. As it isn't clear which next step to take,  they depend on the data team to pull out trends/insights to work on which creates a long feedback loop. 
- **Slow** —  Dashboards require manual analysis to understand what triggers changes. Hence numerous factors and combinations of factors need to be checked to catch a meaningful trend, which becomes extremely time-consuming.
- **Hidden insights** - This problem is two folds. Firstly, dashboards do not show what’s happening underneath the hood. Meaning, if there are two factors canceling out one another, your metric will still appear stable. 
Secondly, it is not possible to test all the different hypotheses at hand manually, analysts tend to go after the usual suspects. Your unknown-unknowns remain never tested. 
- **Waste of resources** — 2-4 hours of manual analysis might appear as part of the job, but throughout the year, it takes up 3-6 months of your team’s time. When the work is repetitive and can be done accurately and more comprehensively by machines, it brings a lot more value to free up your analysts for strategic tasks.  

## Ideal workflows
Dashboards rarely provide actionable insights. Teams should focus on complementing reporting dashboards with analytical products that enable comprehensive diagnostic analytics at the speed of business.

![Dashboard vs Kausa](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Frame_513_998b052725.png)

(image by Kausa — dashboards for reporting and high-level trends; augmented analytics for diagnostic analytics)

New requirements push for solutions based on 5 pillars:
- **Augmented:** leverage Machine Learning to run statistical tests, identify anomalies, and detect relationships between metrics and segments/dimensions with the highest contribution (more detail on augmenting workflows)
- **Proactive:** Business should not have to check 20 dashboards every day to assess whether it’s business as usual. Data teams should not be waiting for business teams to submit requests. Leverage ML to deliver automated insights on key business metrics, covering what’s happening and why.
- **Insightful:** go beyond alerting or showing what’s happening. Provide insights into why it’s happening (read more about the diagnostic analytics gap)
Integrated: leverage already existing workflows and daily tools as the main interface (e.g., Slack). Reduce attrition by not forcing business stakeholders to use multiple platforms/UIs.
- **Conversational:** provide a natural language interface where questions and insights are available in plain English. 

![Screenshot 2023-01-27 at 11.12.44.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2023_01_27_at_11_12_44_37791fd8e8.png)
(image by Kausa - new requirements of Dashboards)


## Use cases

Dashboards need to be complemented whenever speed and/or comprehensiveness of the analysis is paramount. A couple of real use cases that have been rising:

**Automated insights messages on key business metrics drivers**

Many business teams take daily/weekly decisions to optimize the KPIs they own. For instance, performance marketing managers adjust their campaigns and bids on a daily/weekly basis to optimize ROAS (Return on ad spend). With dashboards-based workflows, these teams either spend hours slicing and dicing on dashboards or cut corners and miss out on insights. Augmented analytics can automatically test all the hypotheses, rank the drivers and deliver automatic messages on key business metrics. They can be delivered over Slack, Teams, or Email, making analytics more accessible.

![Product - ahead of game.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Product_ahead_of_game_87220bac96.png)
(image by Kausa - delivering data-driven insights over Slack messages)

**Comprehensive exploration into drivers of key business metrics**

While business teams enjoy easy-to-consume insights, data teams need a smarter way to explore data. The need for diving deeper into the analysis and contextualizing it will never disappear, but it is critical for tools to bring focus to the analysis. So that teams know exactly where to start and focus on segments/subsegments which are prioritized after automated analysis. In the example below, augmented analytics can show that one specific marketing campaign (across hundreds) in Germany is driving a significant increase in AOV for younger customers. Finding these insights in a manual way is like looking for a needle in a haystack. Through augmentation, it takes just a couple of clicks (read more in [this article](https://www.kausa.ai/blog/analysts_dilemma)).

![Kausa ranking.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Kausa_ranking_5a3fe70e45.png)
(Image by Kausa - Priotization of drivers by impact)

## Conquering the last mile of the analytics journey

As businesses collect more data and change faster than ever, BI expectations and needs have evolved. Data teams need to deliver effective self-service to scale impact and outcomes. Actionability requires insightful information delivered in a proactive and easy-to-consume manner. Thus BI needs to become augmented, proactive, insightful, integrated, and conversational. This is the only way to improve speed to insight, make analytics more accessible and multiply the outcome of analytics teams.


Talk to us

Want to make sure your organization is ready for the

new requirements of business intelligence?

## About the author and Kausa

[João Sousa](https://www.linkedin.com/in/joaoantoniosousa/) is the Director of Growth at Kausa. Stay tuned for more posts on how tomore posts on how to nail diagnostic analytics and increase the value of data.

Using machine learning, Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

About the author

Yagmur Anis

Senior Marketing Manager

How automated insight messages can help you unlock the power of proactive analytics

Get an efficiency boost with automated insight messages

Are you tired of receiving rushed messages, asking why one of your key metrics tanked, and leaving you wondering what happened overnight? We've all been there. But what if you could take proactive action before things go wrong?

We know that as data becomes more complex, it can be a real challenge to understand the "why" behind changes in key metrics. With Kausa's ML-powered diagnostic analytics platform, you can analyze your data 24/7 and receive automated insight messages over Slack (or any other messenger platform of your choice) daily. This way you can skip 90% of the manual analysis, knowing exactly where to look for the main drivers of metric changes. This would allow you to focus on contextualizing surfaced data points, taking action to improve metrics, and achieving strategic goals.

## What are the best metrics to follow with automated insight messages?
Kausa can analyze any metric and send you automated messages on changes. But we particularly recommend setting automated insight messages for the metrics that are critical to your business and can change rapidly. Examples include conversion rate, average order value, and return on ad spend across different regions or territories. 

## How does it benefit business and data teams?
In addition to offering a proactive approach to analytics for both business and data teams, automated insight messages also offer unique benefits for each team. 

It helps **_data teams_** to:

- break free from the ticketing system dynamic, by providing effective self-service analytics
- save hours of time by starting their analysis from a place where every hypothesis is tested and the main drivers ranked

It offers **_business teams_**:
an easy way to consume insights, integrated into their daily workflow
reduced time to insights by removing the back-and-forth they have with data teams


## How does it work?
Let’s consider a specific example. You are part of the marketing team of an eCommerce platform and weekly active users (WAU) is one of your Northstar metrics. Instead of you logging on to a dashboard every day to view how WAU changes, Kausa pushes you daily notifications on Slack, where you already collaborate with your teams. 

When everything behaves as expected, you can take a quick glimpse at this message and get back to your other priorities. And when there is a significant decrease in your weekly active users(WAU), Kausa points out the customer segment that is contributing the most to this change. For example, highlighting that even though there was a drop in the total number of users that churned, there was a sharp decrease in the segment of users with a returning streak of 4.

![Product - ahead of game.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Product_ahead_of_game_87220bac96.png)

This simple prompt allows you to skip hours of menial work on dashboards, and instead, direct your attention to what is happening with returning users. So by the time the question of “Why our WAU is down” comes up, you would be already looking into what has triggered a negative trend in returning users. You can focus on identifying what stopped them from returning to the site/app for purchase and take action to recapture their attention before it is too late.

Setting up automated insight messages with Kausa is easy. They are fully customizable and you can choose the messages to be delivered to public or private chat channels. Once you receive a notification, you can collaborate directly with your team on the same platform, and click to explore further in Kausa.

Businesses that change fast need proactive action to succeed. As a result, companies that reduce their time to actionable insights and catch opportunities at the right time will be the ones to stay ahead of the curve. 


Try out Kausa today to fast-forward the manual analysis

# About the author and Kausa

[Yagmur Anis](https://www.linkedin.com/in/yagmur-anis-70311120/) is the Sr. Marketing Manager at Kausa.

Using machine learning, Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

About the author and Kausa

Will analytics teams conquer the last mile of analytics in 2023?

What’s next for analytics in 2023?

I’m overenthusiastic about the end-of-the-year wrap-ups and predictions. It’s the perfect time to zoom out from the daily grinding and see the big picture. I love reading articles on learnings and trends from my peers as well as industry leaders, taking time to review my year, and planning for the next one. So I wanted to share my take on what will be top of mind in data analytics next year. 

2022 hasn’t been the easiest year for most. Due to the recession, we witnessed a high number of layoffs and budget cuts. “Doing more with less” became a phrase that we all used more and more. And I expect it to be the motto of 2023. 

So, I predict an increased focus on efficiency, business value, and maximizing ROI for analytics teams. As a result, only projects and technologies that deliver the following will be prioritized:
- help teams save time and resources, increasing efficiency
- help teams save revenue and cut costs, maximizing data & analytics ROI
Here are my predictions for some of the most important analytics trends heading into next year (in no particular order).

&nbsp;

## Prediction #1 — focus on proving the ROI of analytics
Data-forward companies made significant investments in technology and people over the last years. Initially, their focus has been on collecting, storing, managing, transforming, and displaying data to establish a strong core. 

Data quality is essential to creating meaningful results, but it isn’t enough to create business value. If we were to see the data analytics journey as a marathon and the business impact as the medal, delivering actionable insights that inform daily and strategic decisions are a must-have to complete the race. Hence this is where the leaders are shifting their focus to maximize the ROI. 

So how can you get there?
- Closer collaboration with the business teams (e.g., nailing priority use cases together, and daily/weekly performance reviews to review and share insights)
- Speed and comprehensiveness in analysis to truly empower decision-making
- A strong focus on measuring business value and continuous iterations to achieve the best results

&nbsp;

## Prediction #2 — most decisions will be augmented with ML

To be able to keep up with the pace of data and business, the “doing more with less” motto surfaces once again. Businesses need to augment workflows and automate menial tasks, accelerate speed to insights and break out of speed and comprehensiveness trade-off to create true business value and maximize the ROI of data analytics.

Decisions that use data can be automated in a variety of ways and fall somewhere between being mostly human-based and entirely automated. Gartner predicts that by 2025, 95% of decisions that currently use data will be at least partially automated. In the age of artificial intelligence, organizations that embrace some degree of decision automation are likely to achieve competitive advantage through more rapid, sophisticated, and granular decision-making. 

In BI, most decisions are going to be augmented soon. Two main use cases I see emerging in most data-forward companies are:
- Human-in-the-loop diagnostic analytics: using ML to test every hypothesis in the data and score the drivers of key business metrics changes. Humans can focus on the last mile of filtering out the most relevant ones and making decisions
- Automated alerts on key business metrics changes that are delivered via Slack/Teams in natural language

&nbsp;

## Prediction #3 — data & analytics roles will further specialize

As professions start to mature and become more complex, further specialization often takes place. This has been a natural evolution and analytics won’t be an exception. 

Currently, data & analytics team roles are segmented primarily based on the stages they own in the data analytics workflows
- Data engineers pipe the data in
- Analytical engineers clean the data
- Data analysts and scientists draw insights from the data

These core roles will stay, but there will be further segmentation focused on specific business problems and goals.
- **The analytics engineer** role will continue to grow, ensuring data modeling and delivering clean datasets to data analysts and business users. This trend also frees up data analysts.
- **Data analysts** will become closer to the business, focusing solely on uncovering insights and improving decision-making to become trusted advisors and maximize business impact
- **The data product manager** role will continue to grow, boosting the adoption and monetization of data products
- **The analytics translator** role will continue to grow as a way to close the data literacy gap and maximize the value of analytics use cases

&nbsp;

## Prediction #4 — data mesh for data democratization and scalability

Data mesh has been the hype word of the data talk for some months. But what is triggering this trend? The desire for democratization and scalability of the data architecture. 

Many organizations are still struggling since their architecture choices were driven by technology instead of business needs. Data mesh aims to solve this problem by federating data ownership within the organization and focusing on efficiency, innovation, and transparency. 

It creates a distributed, decentralized way of dealing with data, making it faster to share and create data products. Data products are owned by independent cross-functional teams but abide by central governance to ensure interoperability and consistency. 

Breaking down silos, improving the reuse of data, and fueling innovation, data mesh and business problem-oriented architecture will be the way to go for companies that want to do more with their data. 

&nbsp;

## Prediction #5 — analytics teams define speed to actionable insight as a north-star metric

Most of the data-driven answers take days or even weeks. As companies now have data available in near real-time, decision latency is the current bottleneck in analytics. As businesses change faster than ever, these answers are not coming in at the speed of business. This reactive approach has significant costs: direct negative business impact, undermined data culture, and a downward spiral leading to attrition.

Improving this situation is becoming a top priority for many analytics teams. The top analytics teams have defined speed to actionable insight as their north star metric to ensure actionability, maximize business impact and elevate their data culture.

Speed to actionable insight is a simple concept: the time from a business question to a data-driven decision or action. It already includes the multiple iterations that are often required. This way analytics teams not only focus on answering questions but rather proactively work with the business to drive actionable insights.

This is driving innovations like augmented business root-cause analysis and automatic alerts and insights messages powered by ML. This ensures that teams leverage all the available data, proactively delivering analytics at the speed of business.

&nbsp;

## Prediction #6 — Data regulations will become more top of mind

Regulations such as European GDPR, Canadian PIPEDA, and Chinese PIPL pushed companies to rethink the way they collect and work with data. I - and also analyst houses such as Gartner- expect to see more countries introducing new rules and legislations in 2023. 

So it means that there is more work to be done to win the trust of customers so that they will be more willing to share their information. To achieve this, businesses need to better structure, record, and create more transparency on how they process and handle data. Moreover, effective data governance will become more relevant. Depending on the compliance stage, this could mean some retrospective work on the data at hand, how it has been collected and stored and how it has been used. 

Perhaps not the most exciting prediction on the list but a critical one in maximizing the value of data, adhering to security standards, and preventing any violations and fines. 

&nbsp;

## Conclusion

Teams keeping a strong focus on ROI and not letting any resource or minute go to waste will be the winners. I hope to see analytics teams working closely with the business and focusing on data analysis, insights generation, and taking action to maximize the business value. I expect data foundational projects to be deprioritized unless they are critical or bring significant cost savings. 


Is increasing efficiency and business value through analytics a priority for you?

Kausa can help you maximize the ROI data and analytics. 

[João Sousa](https://www.linkedin.com/in/joaoantoniosousa/) is the Director of Growth at Kausa. Stay tuned for more posts on how tomore posts on how to nail diagnostic analytics and increase the value of data.

Using machine learning, Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

Data analysts should become trusted advisors powered by data

It’s about time we elevate the data analyst role

_NOTE: Most data & analytics roles are ill-defined. This article focuses on the data analyst role, which is often described as a BI analyst. Depending on the organization, there’s also a bit of overlap between the business analyst and the data scientist role._

The role of data analyst has rapidly evolved over the last few years. With the explosion of data complexity and business expectations, they face many challenges. 

In the earlier years and in companies with lower analytical maturity, data analyst has been a generalist role. It has been responsible for almost the whole data and analytics marathon. Starting from preparation and visualization, and all the way to analysis and insight communication. 

The growth and innovation in the data analytics space changed the meaning of being data-driven and the requirements for the data analyst role. This created the need for specialist roles and a strong emphasis on the last mile. This way data analysts are more focused on analysis, insights, and decision-making.
![Evolution of data analyst.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Evolution_of_data_analyst_4acdc18f3f.svg)

In line with this idea, Cassie Kozyrkov defines the key role of the data analysts as “looking up facts and producing inspiration for you” and puts a focus on the demanding nature/pace of the work by stating “The analytics game is all about optimizing inspiration-per-minute.” 

I usually define this concept as speed to actionable insights. 

Speed to actionable insights plays a key role in increasing the value of analytics and helping data deliver on its promise. Hence it should be defined as a north star metric for data analysts (Read more on accelerating speed to insights [here](https://medium.com/r/?url=https%3A%2F%2Fwww.kausa.ai%2Fblog%2Fcost-of-reactive-analytics).)

However, most companies are far from there, due to various challenges. Let’s look into what’s holding data analysts back.

&nbsp;  

## Challenges of data analysts

Data complexity and business expectations have both risen. At the same time, traditional BI tools haven’t evolved much in the last 20 years, creating a gap in many data teams. Here are the main challenges faced by today’s data analysts:

### Ill-defined role
There’s a general problem with role delineation in data & analytics. It’s not only specific to data analysts. Its role has a significant overlap with data scientists and BI analysts (and sometimes also business analysts).

![Ill-defined data analytics role.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Ill_defined_data_analytics_role_532520df46.svg)

If you look at most job descriptions, they will highlight that the mission of data analysts is to deliver insights. This is very ambiguous in itself — is the role to be a self-service enabler (i.e., answer to business requests and build dashboards) or to deliver recommendations? Organizations have different visions for the role of a data analyst, varying from self-service enabler (i.e., answer to business requests and build out dashboards) to trusted business advisor powered by data.

### Value misperception

Data analysts are often seen as “second-class citizens” and feel left behind in their “technical” expertise by their Data Science counterparts. The data science role is associated with higher compensation and status, contributing to this gap. 

### Most time spent on fire-fighting and low-value-added tasks

Data analysts spend most of their time on low-value-added tasks, often in a very reactive approach (e.g., fixing broken dashboards). Many operate mainly as “dashboards/report factories” and are very disconnected from the business.
 
&nbsp;  

## Status quo

These challenges are holding data & analytics teams back, as analysts produce a lot of output, but very few business outcomes. In practice, most data analysts spend most of their time on low-value-added tasks, mainly focused on descriptive analytics, only presenting what happened.

Here’s a breakdown of the 3 most common data analyst states I’ve seen:
![States of data analyst.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/States_of_data_analyst_94771fc2da.svg)
### State I — The data wrangler

You can think about “the Data Wrangler” as a data analyst stuck in the initial generalist definition of the data analyst role. S/he is responsible for doing anything and everything data related ranging from pulling data from different resources, and reports, cleaning the data and preparing dashboards. Therefore they barely have time to focus on surfacing insights, and the value they deliver is diluted.

_Outcome:_ Teams are stuck in the “what”, only able to identify what changed in the business performance and unable to answer “why”. The value of analytics is unclear causing a weak data culture. 

### State II — The dashboard builder

While the “Dashboard builder” state is a bit more evolved than the “data wrangler”, these analysts are not focused on actionable insights. They spend most of their days on visualization,  building, and managing dashboards. They look for insights reactively to answer the questions of the business teams or to put out fires. 

_Outcome:_ The value of the analyst remains uncovered. There is a significant bias in the decision-making process preventing the discovery of real insights, and the analyst teams are overwhelmed with ad hoc requests. 

### State III — Analyzer
This is a mature state where the role of the analyst is better defined and the data culture of the company is already strong. Analysts’ focus is on delivering insights but they are unable to uncover them at the pace of business.  They often also have a lot of non-analyst work to deal with. There is a need to augment analytics workflows to free them up and enable them to focus mostly on the last mile of analytics.

_Outcome:_ Too much time spent on repetitive tasks (e.g., slicing and dicing on dashboards) hinders speed to actionable insight and business impact leading to missed opportunities

&nbsp;  

## The future data analyst — business advisor
Working together with some of the most data-forward data companies, I witnessed the business impact that data analysts can deliver. These advancements are fundamentally changing how a data analyst functions within the company. They attend weekly business meetings to present what happened, why did it happen, and the so-what, sharing proactive solutions for the business.

A Data Analyst’s primary goal is to surf vast datasets quickly, liaise with business stakeholders, and surface potential insights. Speed is their highest virtue. 

The result: the company gets a finger on its pulse and uncovers real insights at the speed of business. This generates the inspiration for decision-makers to select the most valuable quests for Data Scientists. I’ve identified three main characteristics that make data analysts trusted business advisors:

- **Partnership** — best-in-class teams have their data analysts embedded into business functions and partner very closely with business stakeholders.
- **Proactivity** — data analysts share proactive business recommendations by spending significant time on diagnostic analytics instead of just showing/describing what happened. They come for daily/weekly/monthly business reviews with causes as to why key metrics are changing and potential recommendations.
- **Domain/business expertise** — understanding of the business and domain (e.g., product) to connect the dots and drive actionable insights

&nbsp;  

### What it takes

This transformation needs to be enabled by augmentation, coupled with a culture and people shift. These are the main requirements:

- **Augmentation** — this is the main enabler behind this shift. Delivering actionable insights at the speed of business requires augmentation of traditional BI workflows to eliminate the speed — comprehensiveness trade-off ([see more here](https://www.kausa.ai/blog/analysts_dilemma)). Moreover, to become trusted advisors, analysts should not be spending hours slicing and dicing on dashboards. Instead, they should focus on finding real insights, by leveraging ML to run statistical tests and pointing out business teams where to look, and closely partner with them. This is the only way to conquer the last mile of analytics. derive insights, communicate them properly and drive action. 
- **Business and domain expertise** — data analysts need a good understanding of the business and domain expertise they’re working on. Thus it’s getting more common for analysts to specialize in one business function (e.g., marketing analyst) and be embedded into the corresponding function (e.g., marketing team) (Read more about the importance of domain expertise in [this article](https://medium.com/r/?url=https%3A%2F%2Fcounting.substack.com%2Fp%2Fhow-do-we-actually-pull-stories-out) by Randy Au)
- **Soft Skills** — to become trusted advisors, data analysts need to go beyond technical skills like SQL language. Business acumen, communication, listening skills, curiosity, empathy, and storytelling are critical to becoming strategic partners, developing domain expertise, and driving recommendations and action
- **Data culture** — the right data culture where business stakeholders see data as a key asset and are open to close collaboration with data analysts is a key requirement. Driving this culture requires data teams to gain the trust of their counterparts, by focusing on business value.

&nbsp;  

## How to close the gap
This evolution requires a mindset, skills, and cultural shift enabled by the right processes and tools. Here are a couple of specific best practices:

### Tools — augmented analytics as an enabler
- Augment existing workflows so that analysts get granular insights at the speed of business and have enough time to work on the last mile of the analytics marathon.
### People — develop expertise to make analysts trusted advisors
- Embed your analysts into business functions
- Promote learnings exchange among different analysts (also cross-functional)
- Develop best practices for technical and non-technical matters (e.g., how to conduct root-cause analysis, how to present insights)
### Culture — evolve your data culture to maximize value
- Promote a culture focused on business impact — develop a framework around measuring the value, promote success stories, etc
- Ensure that analysts develop domain & business expertise by providing context, creating alignment
### Processes-support the whole transformation
- Develop a set of processes to facilitate knowledge transfer between business and data teams (e.g., data jams sessions, business deep dives)
- Set up weekly or monthly business reviews performance and involve your analysts

&nbsp;  

## Elevate your data analysts to maximize business impact
It’s about time that data & analytics teams conquer the last mile of the analytics marathon. Companies have invested significant resources in data collection, preparation, and visualization. Now it’s time to focus on the finish line — data analysis, insight communication, and better decision-making. Elevate your data analysts to conquer this last mile and deliver on the expectations (and ROI).


About the Author and Kausa


Why accelerating speed to actionable insights is critical


​​The cost of reactive analytics

The growing complexity and speed of data available to businesses today should make decision-making easier.  While the explosion of data triggered major advances in data engineering, storage, and computing, it also brought new challenges.

There’s been a big wave of innovation focused on reducing data latency. Many companies are able to collect data in real-time but analyzing it for insights still takes days or weeks, while business stakeholders impatiently wait.

**Why is this a problem?**

Business can’t wait. To thrive in a world that is changing faster than ever, business stakeholders need quick answers to performance changes. Data is only valuable when it is actionable, so the analysis and insight creation shouldn't be playing catch up with decision speed but bring agility in the decision-making process.

In this context, a key analytics metric is decision latency.

_**Decision latency**  is the amount of time it takes for a team to make a decision in response to a business change .  In other words, time to actionable insights*._

*An insight is more than an observation. It’s a finding that triggers teams to act. And for it to be worthwhile, it needs to be found before the window of opportunity is lost.  An increase in decision latency causes missed opportunities and creates a setback. (read further [here](https://medium.com/towards-data-science/how-to-deliver-real-data-driven-insights-e94f2b386c18))

&nbsp;  

## Why are most analytics teams so reactive?
Most teams are mainly reactive, missing the window of opportunity to drive business impact. Common symptoms I see that create a snowball effect:

- **NEW CONTEXT:**  With more data available than ever before, data teams find themselves facing a new problem: not knowing where to look in the data and where to focus their attention. 
- **PEOPLE/CULTURE:** Traditionally data and business teams worked separately but the new information age demands close collaboration. Analytics teams that are disconnected from the business lack the domain expertise/business context to identify top priorities and don't know what they should proactively look into. 
- **PROCESS:** Many teams try to implement a “self-serve” model but most fail to surface insights and business teams depend on analytics teams to get them. However, there isn’t a structure/consistent way to collaborate. Analytics teams mainly respond to stakeholders' questions/requests (ticketing systems, Slack messages, etc). 
- **TOOLS:** Teams are working with BI tools that are designed for the data amount and complexity of the past decade/s and losing a lot of time on manual slice and dice. Additionally, it’s still on the data organization to prepare and maintain the data, which often requires custom efforts for each new metric or project.

All leading to a direct negative impact on the business, undermined data culture as well as frustration and attrition.

![snowball v1.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/snowball_v1_188ae81aed.svg)


## Negative consequences of a reactive approach

Time is money. So, a reactive approach has a direct negative business impact. The companies failing to act on opportunities on time are unable to drive outcomes. Moreover, decision latency has a lasting impact on the data culture and often leads to a downward spiral with dire consequences such as attrition.

**1. Direct business impact — The missed window of opportunity**

A proactive approach helps teams prevent negative outcomes and maximize performance. Whereas with a reactive approach, at best you can hope for is to minimize losses. Hence every minute/day the answers aren’t found could mean a decrease in profits/market share or loss of customers.  

One example I’ve heard this week: the active user base was declining for 2 weeks when the growth leader asked the analytics team to look into this. After 1 week, the team found out that a competitor was running an ad campaign for a specific term that was taking away significant installs. By the time the team found these insights, it’s been 3 weeks and the estimated business impact is worth millions! What if the team would proactively explore this after a couple of days?

Imagine how many smaller changes go unnoticed… Furthermore, if your teams are doing what can be automated manually, it is a waste of resources that can be allocated to value-added tasks. 

_**Outcome:**_ significant missed opportunities with a direct impact on the company’s bottom line, losses due to inefficient use of data teams

**2. Undermined data culture**

Delays between business questions and data-driven answers undermine the data culture of many organizations. If insights are often delivered too late, business stakeholders will adjust their expectations. In practice, they will start neglecting data inputs and start making decisions without data.

_**Outcome:**_ undermined data culture as stakeholders don’t expect to have data insights at the speed of business

**3. Downward spiral leading to attrition**

A reactive data team usually falls into a downward spiral:

- Ill-defined requests and constant follow-ups (e.g., the typical “quick question” over Slack)
- The team is overloaded with so many requests
- The data team is detached from the business and doesn’t see the impact of their work
- Highly dissatisfied and overworked teams

_**Outcome:**_  downward spiral affecting analytics team’s motivation and performance, potentially leading to attrition

&nbsp;  

## Ideal scenario

Best-in-class analytics teams proactively influence business decisions on a daily/weekly basis. They don’t wait for questions or requests from business stakeholders. Instead, they are strong business partners and proactively share relevant insights and recommendations.

Maximizing the value of data requires proactivity. And not taking action till you observe a sharp decline in your dashboards and then spending days on root-cause analysis can be very costly.

Proactive teams can:
- Deliver insights at the speed of business
- Anticipate the questions they might receive and prepare answers beforehand
- Become a trusted business advisor powered by data

As a result, these teams push the boundaries of data culture and literacy, maximizing the business value of data & analytics. A couple of examples I’ve come across throughout my work at Kausa with some of the most data-forward organizations:

- Teams rapidly test and implement new technologies to augment analytics and create agile workflows
- Data and business teams come together for daily/weekly business reviews with specific hypotheses/recommendations as to why things are changing and what to do about it (what + why + so-what)
- Data teams are the ones to raise questions to the business (e.g., I’ve noticed this is happening for 3 days, I looked into the data, and… what do you think?)

&nbsp;  

## How to accelerate speed to actionable insight
A cultural shift and backing from leadership are critical for this transformation. Given that this challenge is triggered by technology, technology also holds the key to overcoming this challenge. Augmenting analytics workflows can speed up manual slice and dice by ten times and help teams to get out of reactive work cycles by offering comprehensiveness and speed at the same time. 

![proactive analytics v1.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/proactive_analytics_v1_f6aaefba0d.svg)

**Promote a cultural and mindset shift**

Data teams need to partner closely with business stakeholders to: 
- Define business goals and priorities together
- Define speed to actionable insight as a north star metric - e.g. setting goals like answering 95% of the questions within one hour or accelerate time to actionable insight by 75%
- Promote a data culture where teams don’t wait for questions/requests from the business
- Reduce fire-fighting by developing good processes, automating as much as possible, and preventing issues.

**Augment traditional workflows**

Achieving speed and comprehensiveness together is key to taking proactive action. Hence teams need to:
- Leverage augmented analytics solutions that use ML to augment current workflows and automate repetitive tasks
- Test all the possible drivers affecting metrics daily and develop alerts when something unexpected happens
(Read more on how to augment analytics workflows [here](https://www.kausa.ai/blog/analysts_dilemma))

**Embrace data products**

Teams can trigger a shift from Data as a service to Data as a product mindset by:
- Managing all data assets as products
- Embracing a decision intelligence product for diagnostic analytics vs approaching in an ad-hoc/reactive manner to reduce inter-team dependencies and establish a real self-service culture. 

&nbsp;  

## In a nutshell

A reactive approach to analytics can damage your business both monetarily and culturally. To avoid and improve:

**Watch out for the symptoms:**
- Teams losing focus and growing gap between data and business teams
- Inefficient processes and treating data teams as a ticketing system
- The long time between questions and data-driven answers
- Teams don't share any insights proactively, but rather just react to requests and questions

**Communicate the consequences of reactive analytics:**
- Negative impact on the business
- Undermined data culture
- Increase in attrition

**Start accelerating time to actionable insights by:**
- Promote a cultural and mindset shift
- Augment traditional workflows
- Embrace data products
- Create leadership support to drive a cultural/mindset shift








Do you want to break the vicious cycle of reactive analytics?

Kausa can help you accelerate time to actionable insights by augmenting diagnostic analytics workflows. 

## About the author and Kausa

[João Sousa](https://www.linkedin.com/in/joaoantoniosousa/) is the Director of Growth and [Yagmur Anis](https://www.linkedin.com/in/yagmur-anis-70311120/) is the Sr. Marketing Manager at Kausa. Stay tuned for more posts on how tomore posts on how to nail diagnostic analytics and increase the value of data.

Using machine learning, Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

Why accelerating speed to actionable insights is critical

Proactively delivering real insights into metrics changes


Diagnostic analytics — how to conduct a root-cause analysis

_Note: The term root-cause analysis is commonly used in IT and data engineering as the process to identify root causes of faults or problems. This article focuses on diagnostic analytics to understand the drivers of business metrics changes._

Why is revenue down? Why did the conversion rate spike? Why is the average order value flat?

Depending on your industry and business goals, your key metrics might be different. But if you care about answering “How do I improve my key metrics?”, you need to understand why they changed first.

I see many teams struggling to answer to identify the root causes behind business metric changes. Depending on their diagnostic analytics maturity level, the reasons vary (see [diagnostic analytics gap](https://www.kausa.ai/blog/diagnostic-analytics-gap)).

## Why is this relevant?
To be able to come up with recommendations to improve performance, teams first need to understand the changes happening in their metrics. Without looking under the hood it is not possible to find true insights -not just observations- for teams to act on (see the article “[How to deliver true data-driven insights](https://towardsdatascience.com/how-to-deliver-real-data-driven-insights-e94f2b386c18)”).

Ideally, they should do this regularly as opposed to investigating when a sharp decline or spike is observed in metrics. Otherwise, teams will get stuck in reactive analytics loops which could be quite costly due to delayed insights and actions.

## Status quo
Based on conversations and collaborations with dozens of data & business teams, one can group diagnostic analytics maturity under 4 states, depending on the approach to diagnostic analytics, analysis thoroughness, and time to insight.

![4 States of diagnostic analytics.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/4_States_of_diagnostic_analytics_b03f540850.svg)
(States of Diagnostic Analytics — Image by Kausa)

### Stuck in the what
Teams are mainly describing what’s happening (e.g., the metric is going down) and connecting the dots with high-level qualitative facts (i.e., events that happened in the business like website updates). The process is unstructured and not data-driven. They are mostly in firefighting mode, looking into the ad hoc requests coming from the business teams.

_Outcome:_ Untapped potential that teams are not even aware of, weakened data culture

### The usual suspects (i.e., “business-led” approach)
Teams have a very high bias towards only testing the usual suspects or the hypotheses proposed by the business teams. While it is useful to receive context and direction from the business units, teams should go beyond just testing these hypotheses. This can introduce significant bias and make teams overlook valuable opportunities.

_Outcome:_ Significant bias in the decision-making process, leading to missed opportunities. Lack of real insights.

### Need for speed
Teams are in a more mature state of diagnostic analytics and they recognize the value of drilling down to the why. While they have a more structured approach, performing comprehensive root-cause analysis with existing workflows is too complex and time-consuming. Often they do not get to the insights as fast as the business demands.

_Outcome:_ Insights are uncovered too slowly to make a true business impact and improve decision-making.

### Ideal state — the full force
The full-force diagnostic analytics requires a proactive approach based on an impact-oriented comprehensive root-cause analysis.

![Screenshot 2022-12-01 at 17.31.11.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_12_01_at_17_31_11_cead1d05b0.png)
(Slide by the author — Reaching full force in diagnostic analytics)

**Fast AND Comprehensive**
Teams use machine learning to augment diagnostic analytics. This way they can test all the possible drivers behind the changes within minutes, without sacrificing speed or comprehensiveness and removing human bias.

**Proactive**
Teams not only look into drastic/surprising changes but keep a pulse of their metrics on a daily/weekly basis depending on the speed of the business. They go into the performance meetings already equipped with the potential drivers for the changes to drive the conversation. Proactively share these insights daily/weekly instead of reacting to questions from the business.

**Impact-oriented**
A small change in one of your largest markets could be still significant but a major change in a tiny subgroup might appear way more prominent unless you are taking true impact/actual contribution into the account. (see more [here](https://www.kausa.ai/blog/why-analytics-needs-an-impact-driven-approach)).

When looking at a specific subgroup, its contribution to the global metric change has to be derived from 2 effects:

Change in the given metric for this subgroup
Change in the subgroup volume (share of this subgroup over the global population)
These two effects can be calculated individually, assuming ceteris paribus (all other things being equal). By adding up both effects you get the contribution of this subgroup to the global metric change.

Let’s assume you are looking into Average Order Value (AOV) changes week over week.

The Global Average Order Value has increased from 50€ to 52€.

You want to understand the contribution of each country to global change. Let’s take France as an example:

- _**Change in metric:**_ AOV in France increased from 60€ to 70€
- _**Quantity/volume:**_ the share of French orders stayed constant at 10%
- _**Actual impact:**_ derived by the sum of both effects
- _**Change in metric:**_ this increase in the AOV of France orders contributes positively to the global AOV. Assuming ceteris paribus (i.e., the size of France would have remained 10%), the effect is (70€-60€)*10% = 1€
- _**Change in volume:**_ 0 (given that the size remained constant)
Given both effects, if everything else were to stay the same, the global AOV have moved up 1€ (i.e., this is France's contribution to the global AOV change)
The best approach to conducting root-cause analysis
The best practice methodology consists of 3 key steps: narrow down the change to the most impactful subpopulation(s), analyze related/dependent metrics, and connect the dots with actions taken, interactions, and external events. Then it’s critical to ensure those insights are properly communicated and presented to drive actions.

## 1. Narrow down the change to the most impactful subpopulations

Narrow down the change to subpopulations driving most of the change after testing all potentially relevant factors to avoid human bias. This is the part that is humanly impossible to do with manual resources. Even if you are spending days/weeks on a single change, it is highly unlikely that you would look into all the factors and score them according to impact.

Using machine learning / augmented analytics, thousands, even millions of combinations can be checked within minutes and only the significant ones can be scored according to the actual impact explained earlier.

## 2. Analyze related/dependent metrics

After narrowing down the change to the most impactful subpopulations, look at how related/dependent metrics evolved for these subgroups. Oftentimes a metric is a function of other metrics (e.g., ROAS in gaming is a function of expected revenue, cost, and installs). In these cases, having this metrics tree in mind can help further understand why things are changing (e.g., ROAS decrease due to a spike in cost)

## 3. Connect the dots

Collaborate with the business teams to connect the dots and drive actionable insights. Brainstorm on which actions were taken, as well as interactions, events, and external factors that can influence the business metric you’re looking into. Relevant facts can be segmented into 3 groups:

- **Internal** (i.e., actions taken by the company)
- **External and industry-specific** — actions that were taken by the competition (e.g., new marketing campaign) or general trends in the industry (e.g., summer seasonality) that affect the business metric you’re looking
- **External and geographical-specific** — broader actions that affect a specific region (e.g., elections in Brazil)

## 4. Present and communicate the insights effectively

Explain key facts and drivers upfront, stating your hypotheses and quantifying the impact. Develop consistent ways to present results (e.g., waterfall charts, memos like the one below)

**Example**

Let’s go back to the Average Order Value example — AOV

AOV tracks the average dollar amount spent each time a customer places an order on a website or mobile app. Product/Marketplace teams aim to maximize AOV by testing different campaigns and making changes on the website/apps over time.

In this case, AOV increased from 62.4€ to 63.7€ WoW. This is a 2.1% increase.

![01 metric change snapshot.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/01_metric_change_snapshot_e3d7c90d90.png)
(image by Kausa — metric change snapshot)

As it isn’t a big spike or a drop, it would be pretty likely for many teams to ignore this. At the full-force level, every week key business metrics like AOV are under examination to uncover every opportunity and maximize business impact.

So, where is this 2% coming from?

By augmenting the workflow, you could test all the factors affecting AOV — more than 500.000 combinations for this case — and see which drivers have the biggest impact.

![02 prioritization of main drivers.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/02_prioritization_of_main_drivers_4aa1fcb090.png)
(image by Kausa — prioritization of main drivers)

Within seconds, I can tell that country, campaign, and customer age are the biggest impact drivers. And interestingly, there are quite a bit of factors counter-affecting one another and canceling each other out in a high-level view.

Let’s see which countries are contributing negatively and positively.

![03 prioritization by country.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/03_prioritization_by_country_d6d7c4da87.png)
(Image by Kausa — actual contribution by country)

Interestingly, Germany is performing really well while France, the US, and Korea are not so much.

You’re curious why Germany is performing particularly better compared to other countries. So you drill down a bit more to identify that Campaign ID is the subfactor that contributes to the AOV increase in Germany. But which campaign/s?

![04 main drivers Germany.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/04_main_drivers_Germany_871bb5720b.png)
(Image by Kausa — main drivers prioritized for Germany based on actual contribution)

There is one campaign in Germany that contributes 3.18€ to the overall AOV. Interesting…

![05 campaigns in Germany.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/05_campaigns_in_Germany_2d62e0f779.png)
(Image by Kausa — Campaigns in Germany prioritized based on actual contribution)

Now you look into related metrics to gain more context. It looks like the marketing team increased the ad spend for this campaign, creating a boost in both order value and order volume. The volume of the subgroup is rather small. You might want to inform the teams we can test in other segments too.

![06 viewing relevant metrics.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/06_viewing_relevant_metrics_85a6be7a38.png)
(Image by Kausa — Viewing relevant metrics)

Before you connect with marketing teams, you want to see if you can gather any other insights. Age seems to be an important subfactor. The campaign is performing especially well for the 18–20 age group.

![07-three dimensions.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/07_three_dimensions_d95d625e86.png)
(Image by Kausa — Looking into three dimensions together)

Great! Now you are ready to communicate these findings with the marketing team to gain more business context.

After a quick check-in, you find out that they are testing new creative visuals for this campaign. With this information in hand, marketing teams start testing the campaign in other regions and data teams will be on the lookout to see if the same trend is observed in other countries. And you can allocate time for more interesting data analytics projects instead of spending hours/days on drilling down.

Win-win.

## Bottom line: Impact-driven approach
Looking at the status quo, there is a big opportunity to improve diagnostic analytics and maximize business value. Currently, the way that diagnostic analytics is handled can sometimes feel like mayhem (from constant firefighting to missed deadlines, etc…). But it is possible to bring logic and structure into this madness. Where to get started? Depends on your starting point but consider the following:

- Evolve your modern data stack to include a decision intelligence/diagnostic analytics platform to be able to augment the workflows
- Start testing all the hypotheses at hand to surface overlooked opportunities
- Score the drivers affecting your metrics based on their actual impact to be able to focus on what truly matters
- Communicate the findings with relevant stakeholders proactively




Read Diagnostic Analytics Gap article

Are you interested in maximizing business value using diagnostic analytics?

Focusing on speed, comprehensiveness and, impact all at once, we can help you reach the full force. 

## About the author and Kausa

[João Sousa](https://www.linkedin.com/in/joaoantoniosousa/) is the Director of Growth at Kausa. Stay tuned for more posts on how tomore posts on how to nail diagnostic analytics and increase the value of data.

Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance, using machine learning. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

How a focus on Business and Decision Intelligence can unlock greater business value

Evolve your Modern Data Stack

The journey from raw data to valuable, actionable insights is a long one, and not for the faint of heart. Each step of the journey requires time and effort, and more often than not, different tools. 

The first step – data collection – for example, can already be a daunting uphill battle as companies have data scattered across their various source systems and files. The integration of fragmented data into a centralized data warehouse relies on an ELT or ETL process, for which you may need several providers to connect all your sources. The following steps – storage, transformation, visualization, and analysis – all involve data being relocated or transformed into different formats and structures. Needless to say, it’s no easy feat. 

Brent Dykes, [in his Forbes article](https://www.forbes.com/sites/brentdykes/2022/01/12/data-analytics-marathon-why-your-organization-must-focus-on-the-finish/?sh=72aba1373c3b), refers to this process as a **Data Analytics Marathon**. The grueling nature of this “marathon” is why the data world has come up with so many evolving concepts, philosophies, and tools, to efficiently take on this multi-stage process with reduced effort and hassle. The Modern Data Stack is a set of cloud-native tools that are designed to streamline each step of the way.

Here’s a snapshot of the Data Analytics Marathon: 
![MDS and analytics marathon.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/MDS_and_analytics_marathon_0291951e38.svg)

As Dykes argues in his article, however, and as with any marathon, there’s not nearly enough emphasis on the “last mile.” Modern companies seem to be powering through and losing steam, rather than keeping a steady pace to gear up for a strong finish. If you think about it, the last mile is really the only part where it all counts – in data, it’s when your company can finally take action on the insights generated from data. If you don’t finish the marathon, did the rest of the journey even matter? If data is visualized, but no insights are taken from it, the rest of the process isn’t worth the investment.

To ensure that companies are getting value out of their data investments, the Modern Data Stack is evolving to favor the “last mile” – consisting of BI (business intelligence), DI (decision intelligence), and the human-led tasks of communicating insights across the organization and taking strategic action. Business intelligence allows data to be illustrated and visualized in a way that reveals trends and information. Decision intelligence takes it a step further by helping companies automate manual analysis and diagnose the drivers of the changes within their business performance. Using the insights uncovered and re-allocating the time gained, teams can take actions that will bring business value. 

Let’s expand further on what’s behind BI & DI, and how to finish strong. 

## Business Intelligence (BI)

### Data modeling

Your data is replicated from your data warehouse to your BI platform, and is ready to go. Ideally, your BI platform will have a modeling layer, as well as visualization capabilities. BI platforms like [Whaly](https://whaly.io/) are becoming more robust, offering more than just the visualization component. 

The modeling layer is where you begin to make your data comprehensible to all, since your end business users will need to leverage it for their business decisions. In analytics, modeling is the process of reshaping data into a format that’s easy to understand and to query for your non-technical users. Models are designed to make sense in the context of company definitions and business logic, written through SQL. In many cases, models will be the foundation of your Explorations. 

In general, there’s a higher level of data-savviness required in this step, which means it’s usually handled by data analysts or data engineers. The data team is also responsible for creating a semantic layer by defining metrics and dimensions across the company. 

### Self-service data exploration and visualization

Now that you have your data models and the semantic layer in your BI platform, it’s time to explore and visualize the data in dashboards and charts. This will shed light on what’s going on in your business. 

The tricky part here is that the end users who need to consume the data to drive their business decisions generally don’t come from technical backgrounds. As a result, data teams get treated like a support team, fielding requests for dashboards all day, and end up being a bottleneck for reporting.

How do data teams and business teams collaborate better? The answer is a self-service BI platform. Within the platform: 
Data teams can model the data, create the semantic layer, and ensure that the data is reliable and well-governed.
After the data teams “tee up” the business teams for success by preparing the data, the business teams should be empowered to answer their own questions and instantly get a picture of what’s happening in their business.

A strong self-service layer means that business teams can create their own visualizations and dashboards, based on the metrics & dimensions that the data teams have assigned. It should be customizable, with a range of charting options, so that the “data story” can be told in the most effective way that can be easily grasped, and minimizes the time to insight. 

When self-service really works, and business teams can autonomously run their own queries, this will boost data adoption and foster company-wide trust in data. 

## Decision Intelligence (DI)

So, you have the dashboards and charts in front of you. You can see things like team performance against objectives, trends over the past few quarters, the upward tick in MRR. That’s great, but what do you do with this information? Perhaps there are a few quick analyses you can run to infer if things are looking up or down, but that’s not going deep enough. These days, you can also leave diagnosing why metrics are changing to machines. 

Decision intelligence platforms like Kausa, test millions of factors and combinations of factors that might be driving the change in your metrics and point your attention to the most significant ones so that you can direct your attention to areas that matter the most. And they can do it about ten times faster compared to manual slicing and dicing. 

Till recently, this category has been overlooked by most of the teams, creating a [diagnostic analytics gap](https://www.kausa.ai/blog/diagnostic-analytics-gap). Not having the necessary tools at hand, most teams have been searching for why metrics change by drilling down on dashboards. While dashboards are very informative in terms of providing a high-level overview of what is happening, doing this analysis on dashboards requires a significant amount of repetitive manual work. Teams in slower-moving industries or dealing with smaller, less complicated data sets can make this work. But for a fast-moving business with complex data, this analysis can take hours or days, leading to a lot of missed opportunities. The next step in BI is to accelerate the speed to actionable insights, and this is the exact focus of Decision Intelligence platforms.  

Depending on the way your teams are structured, decision intelligence platforms can be used directly by user acquisition and marketing teams, or to power data teams to become trusted advisors for the business teams they are working with. It aims to initiate a proactive way of working with data, looking into performance drivers regularly and continuously acting on every opportunity to improve business performance. This gives the teams the opportunity to closely align data-initiatives with business goals as opposed to the reactive approach of chasing the answers to questions that are randomly rising due to the fluctuations in KPIs. 

Similar to dashboards, decision intelligence platforms integrate directly into various data warehouses and enable teams to analyze all the available data using an already existing SQL query. This way instead of just testing for 2-3 scenarios you already suspect, you can uncover unexpected drivers hidden in your data and account for counter-factors, canceling out one another and appearing flat in the BI tools. By augmenting diagnostic analytics using machine learning, decision intelligence puts an end to the trade between speed and comprehensiveness. As a result, it unlocks the last mile of analytics by surfacing actionable insights within minutes and frees teams to focus on communicating and acting on these insights. 

—

At this point, you’ve made it all the way through the “last mile,” starting to deliver real value from your data. Congrats on finishing the marathon! 

In summary, make sure your company has the right mindset and tools in place to effectively get you through Business Intelligence and Decision Intelligence – which is where it all counts. By streamlining the BI & DI steps, you’ll not only get a picture of what’s going on in your data, you’ll also quickly learn why it’s happening, so you can take quick action and grow your business. 



How does Kausa fit into your stack?

If you want to evolve your Modern Data Stack for impact,

learn how Kausa can speed up time to insights and unlock business value.

## About the authors and Kausa 
This piece is a joint collaboration between Whaly & Kausa. Yagmur Anis is the Sr. Marketing Manager at Kausa and Anna Lorentz is the Head of Marketing at Whaly.

If you want to talk more about Decision Intelligence, please contact [Yagmur Anis](https://www.linkedin.com/in/yagmur-anis-70311120/). 

If you want to talk more about Business Intelligence, please reach out to [Anna Lorentz](https://www.linkedin.com/in/anna-lorentz-2a37603b/). 

[Kausa](https://kausa.ai) can analyze all your data using machine learning,, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

[Whaly](https://whaly.io/) is a self-service business intelligence platform that empowers data teams to grow data adoption among business users. Whaly allows companies to make better decisions and drive revenue by reducing the amount of time from raw data to insight, and by removing data teams as a bottleneck for ongoing analytics. 


Evolve your Modern Data Stack for impact

How to resolve it with augmented analytics

The analyst's dilemma - too much data, too little time

For digital-first businesses, the only way to observe and understand their customers and business is through data. And one would think that with the abundant data, companies have at hand, data-driven decisions would be simple. But data has not only increased in size but also in complexity. Nowadays companies collect many data points on engagement, sale, channel etc. So analysis can quickly become a daunting and unfruitful task with traditional BI tools. The good news is, there are also easier tools and processes to facilitate data transformation and visualization. However, the adoption level for these augmented analytics tools/ methods (aka diagnostic analytics powered by machine learning) are still very low.
&nbsp;

&nbsp;

## Why should you care about augmented analytics?
The way that we produce and accumulate data and the limitation of human attention create a big dilemma for analysts - too much data and too little time. To create value and business impact from data, teams need to deliver comprehensive actionable insights at the speed of business. However, given the new context and legacy BI tools, most teams end up trading off between:
- **Speed**  -  not getting answers fast enough to make effective and timely decisions
- **Comprehensiveness** -  missing out on insights as they are just scratching the surface and looking through data at a granular level.
![Screenshot 2022-11-22 at 14.29.44.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_11_22_at_14_29_44_c59ff4c725.png)
(Image by rawpixel.com on Freepik)
&nbsp;

&nbsp;
As a result, actionability suffers. The answers gathered aren't real insights or the insights aren't uncovered fast enough to drive business action.

To make this more concrete, let's take a closer look at the current analytical challenges:

**1. Limited human attention**

I hear these very often:

_"There's so much data and so many metrics, we can't possibly monitor them all."_

_"I have to sift through 15 dashboards every day to assess whether it's business as usual."_

_"It would take us days to go through our tools and answer why in a comprehensive way."_

Data storage and transformation capabilities are not the bottlenecks anymore. However, we've been using the same interface for business intelligence - dashboards. As elegantly put by Herbert A. Simon, the current limitation is human attention:

_"In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: the attention of its recipients. Hence a wealth of information creates a poverty of attention and a need to allocate that attention efficiently among the overabundance of information sources that might consume it"_

As dashboards are unable to direct our attention to where it matters the most, many teams only analyze a portion of the available data, which is one of the most common bad analytics practices ([read the article](https://www.kausa.ai/blog/Six-bad-analytics-habits) to find out the other 5).

**2. Fast-changing context**

Businesses change faster than ever. As a result, metrics change faster than our ability to explain why using traditional dashboards-based workflows. At the same time, companies need to embrace agile decision-making to stay ahead of the competition.

As Dave Girouard stated:

_"Speed is the ultimate weapon in business. All else being equal, the fastest company in any market will win."_
&nbsp;

&nbsp;

## Why dashboards fall short in solving the analyst' dilemma
Dashboards are great for descriptive analytics (i.e., understanding what's happening) and metrics monitoring. They also enable high-level exploration to identify main trends over time. However, they fall short in diagnostic analytics (i.e., why did it happen).

## Missed opportunities

As a result of dashboard-based workflows, businesses are missing out on opportunities, given that decision-making is:
- **Reactive**  -  when something changes, business teams ask the data team to look into it or build ad-hoc dashboards. Teams are reacting to changes.
- **Slow**  -  ticket loops between data and business teams make diagnostic analytics very slow. In addition, performing root-cause analysis in dashboards is very time-consuming.
- **Poorly informed**  -  teams often just scratch the surface testing for just a few scenarios overlooking most, as dashboards don't enable comprehensive root-cause analyses.
&nbsp;

&nbsp;
## The role of AI and ML
Machines are great at performing repetitive tasks and computations at scales. In fact, across many domains, AI is not anymore perceived as a way to compete with workers, but rather to supercharge them.
Machines can easily analyze millions of hypotheses in real time, eliminating repetitive slicing and dicing work that keeps teams waiting for insights. By running statistical tests and bringing an impact-focused view, AI and ML can eliminate human bias and point humans' attention to what's relevant.
Image by Kausa - machines can test these many combinations and filter out what's relevant.
![Kausa - combinations.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Kausa_combinations_0d0e885866.png)
(Image by Kausa - Testing many combinations and filtering what's relevant using Machine Learning)

In the example below, augmented analytics can show that one specific marketing campaign (across hundreds) in Germany is driving a significant increase in AOV for younger customers. Finding these insights in a manual way is like looking for a needle in a haystack.
![Kausa - drivers.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Kausa_drivers_696bd28292.png)
(Image by Kausa - Priotization of drivers by impact)

However, machines don't have the context to interpret the data, connect the dots with what's going on in the business, and share insights. Who can close this gap? Humans. They are the ones who can look at the relevant information, filter what's relevant, and connect the dots to drive action.
&nbsp;

&nbsp;
## Human-in-the-loop proactive analytics
Most data scientists overestimate the ability of AI and ML and underestimate the importance of context and domain expertise.

Machines can run statistical tests and compute many numbers in seconds. However, they can't (yet) interpret the results and connect the dots to what's happening in the business (e.g., actions taken, seasonality, external factors, etc). Humans should decide which insights to focus on, communicate them effectively, and drive action to complete the last mile of analytics.
&nbsp;

&nbsp;
## Where do you stand?
There are multiple applications and benefits of augmented analytics depending on the specific business use case. These can be segmented based on two key dimensions:

- **Speed** -  how fast-changing is the environment and how often are decisions made. For instance, performance marketing is often very fast-changing, as marketing managers take action almost every single day to tweak campaigns/bids and optimize marketing spend. On the other hand, product feature decisions are typically few and data analyses can take longer.
- **Complexity**  -  number of relevant dimensions and related metrics. Some use cases are by nature very complex since they encompass a high number of relevant dimensions and related metrics. For instance, in-game optimization involves very wide and complex datasets, given the high number of events stored. On the other hand, a marketing manager only running a couple of campaigns in a specific channel and country manages very little complexity.

Based on these elements, one can group use cases into the following matrix, basing the names on what it feels like to work on them with traditional approaches.
![Screenshot 2022-11-22 at 14.51.14.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_11_22_at_14_51_14_4f465a127c.png)


### 1. No sweat

These use cases are relatively trivial (i.e., a low number of relevant dimensions and relevant metrics). Teams deal with a low level of complexity and have significant time for data analysis.
An example could be a small e-commerce company operating in a single region and running a couple of marketing campaigns. Actions are taken on a monthly or quarterly basis and it's a slow-moving market.

Most likely traditional BI tools are sufficient for this low level of complexity and slow business dynamic, without making the team sweat.

### 2. Running behind

These use cases are very fast-changing and agility is key. Teams need to make decisions every single day or week. Thus stakeholders need quick comprehensive insights.

A typical use case I'm very familiar with is performance marketing/user acquisition. Oftentimes small teams run many campaigns across various countries, channels, media sources, and platforms, targeting multiple audiences with various messages and creatives. They are trying to optimize ROAS (Return on Ad Spend), by identifying what is working and what is not, fast.

**Application:** Augmented analytics platforms can test every hypothesis in the data and point out which are the most significant subpopulations. This enables teams to quickly identify what's moving the needle and then understand WHY by looking at related metrics (e.g., CPI increase in iOS campaigns from Vungle in the USA).

**Outcome:** Campaign performance is understood in minutes rather than hours or days, and time saved is reallocated to creative and strategic work.
 
### 3. No pain, no gain
 
These use cases are very fast-changing and complex. Teams need to make decisions every day or week. They are under a lot of pressure as there are many potentially relevant dimensions and related metrics to test for at the same time. Using traditional methodologies, they are regularly in fire-fighting mode and need to spend significant time to reach real insights, but their work is usually underappreciated as the analysis takes too long. This bucket corresponds to the hardest use cases such as leveraging game data to shape features, drive engagement and retention, or balance game economy.

**Application:** This group does not have the luxury to compromise between speed and comprehensiveness. And they can actually make the most out of automation. They can use augmented analytics to identify where to drill down within minutes and uncover potential opportunities in a proactive manner.

**Outcome:** With the exploratory analysis uncovering where to look and ensuring teams get actionable insights at the speed of business, the tables will turn. They will transition from a state where they are almost harassed by requests, to trusted advisors anticipating where to focus for the business teams.

### 4. High-stakes

These use cases are very complex, requiring lots of data points, and interactions to be analyzed and tied to big decisions. A potential use case could be web analytics or supporting product development in gaming. While you don't want to make changes to your website or to a game level every day due to implementation times, the changes you make are critical for business performance. Before the teams move on with these decisions very complex data with a very high number of variables are collected. 

Compared to the "no pain, no gain" use case, teams working on these cases have a bit more breathing room due to the slower pace but are still under a lot of pressure due to the high stakes. Due to the complexity of data, even with the slower pace, they are only able to test for the usual suspect, leaving true opportunity discovery to the luck of the draw.

**Application:** Augmented analytics platforms enable teams to leverage every data point by combining multiple sources to uncover actionable insights. Going through this analysis way faster, the team will have time to go beyond just the peaks and drops and can also investigate interesting insights that can help inform the strategy.

**Outcome:** Granular and actionable insights will lead to high-value decisions and improved metrics. By looking into every scenario and not letting any opportunity go wasted, teams will have more confidence to inform these critical decisions and will get a seat at the decision-making table.

## Augmented analytics is the solution
Augmented analytics brings together technologies (such as machine learning), processes, and people, to help humans leverage the speed and power of machine scale while incorporating their own process and context to make better decisions - at speed and scale. This is the way to close the diagnostic analytics gap.


Would you like to break free from the speed-comprehensiveness trade-off? 

Learn how you can use augmented analytics to automate slicing and dicing and maximize speed to insights. 

## About the author and Kausa
[João Sousa](https://www.linkedin.com/in/joaoantoniosousa/) is the Director of Growth and [Yagmur Anis](https://www.linkedin.com/in/yagmur-anis-70311120/) is the Sr. Marketing Manager at Kausa. Stay tuned for more posts on how to nail diagnostic analytics and increase the value of data.


Using machine learning, Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value.

Stefan Dörfelt

Co-founder & CEO

How can UA teams play smarter in a high-stakes game using analytics

The new rules of User Acquisition in Gaming

Things have been good for the gaming industry. While the majority of the industries were hit by the pandemic, the gaming industry hit an all-time high in terms of revenue and player growth, and the [market expanded 26%](https://www.marketwatch.com/story/videogames-are-a-bigger-industry-than-sports-and-movies-combined-thanks-to-the-pandemic-11608654990) between 2019 and 2021. 

But the game is changing…

## Changing market dynamics introduce new challenges

Although the long-term performance of the gaming sector looks strong, the next couple of years pose a unique challenge for the gaming industry, as [analysts predict](https://www.marketwatch.com/story/videogames-are-a-bigger-industry-than-sports-and-movies-combined-thanks-to-the-pandemic-11608654990) the market to decline in 2022 and a recovery starting from 2023.

Combined with economic recession, post-covid gaming fatigue, and a saturated market, it is getting harder to attract new users and retain the existing ones, leading to reduced conversion rates and higher costs for player acquisition.

UA has been top of mind for gaming companies as a main driver of growth. Due to the increased price consciousness among players and the limited time they have, the price of acquiring new users is increasing rapidly and eating away from the profit margins. As a result, the focus is shifting from growth to retention and capturing high-value players. 

While the task is getting harder, UA and marketing teams in the gaming industry are not necessarily receiving additional resources. On the contrary, due to the global economic turmoil, budget cuts are placed, teams are shrinking, and/or hiring is put on hold. 

When there is more to do and fewer resources to get it done with, there is no other option but to get smarter and more efficient.
&nbsp;

&nbsp;

## Challenges pile up on the old ones
![putting-last-pebble-pile-stones.jpg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/putting_last_pebble_pile_stones_4b964e4f6f.jpg)
_(Image by javi_indy on Freepik)_

User acquisition in gaming has never been simple due to the existing structure and industry dynamics. Before the new challenges piled up, UA teams were already tasked with a difficult job due to:

**The fast-changing and competitive market:** From multibillion-dollar giants to niche developers, the gaming industry is vast. With more players coming into play and different platforms and game types rising in popularity, the industry is under constant pressure to innovate. 

**Balancing loyalty and novelty:** One of the most critical challenges of user acquisition is not only to launch a successful game but also to keep it running with a high number of DAU (daily active users) and revenue over an extended period. With so many options available and players on the lookout for novelty, UA requires regular assessment and analytics to optimize ROI.

**Complex analytical landscape:** UA teams analyze multiple campaigns for different games, across geographies, media sources, user groups, and platforms. Testing for every potentially relevant factor and combination of factors can be daunting. 
&nbsp;

&nbsp;

## How can UA teams beat the odds?
As gaming industry veteran [Vineet Vashishta](https://www.linkedin.com/in/vineetvashishta/) said "Disney frequently talks about how not all customers are created equal. They arm experts with granular customer data to guide their UA activities and maximize revenue per customer. Gaming companies have access to rich datasets. The ones that can connect in-game behaviors to buying behaviors with analytics and automate their workflows will beat the odds in maximizing revenue per user and retain high-value players."

### **1. Capture high-value users and boost retention**
![pov-happy-young-man-celebrating-his-victory-online-shooter-game-using-wireless-controller.jpg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/pov_happy_young_man_celebrating_his_victory_online_shooter_game_using_wireless_controller_2f64846c69.jpg)
_(image by DCStudio on Freepik)_

Changing marketing dynamics is pushing UA teams to focus on a wider set of metrics. Growth has been the primary focus in the past couple of years and metrics around new users were the top priority. Now teams are putting more focus on improving lead quality, profitability, and retention, making metrics like D30, D90 retention, and CLV more important. 
 
**What to do?**
- Identify the segments of high-value players early on
- Run re-engagement campaigns alongside acquisition campaigns
- Work more closely with game developers and product managers to create more engaging experiences and share insights from performance analytics among teams
 
**How can analytics move the needle?**

Drill down to the why in a fast and comprehensive way. Go deeper and analyze how sub-factors, and combinations of sub-factors affect your metrics and cross-check other metrics to gain context. This can help identify high-value players and the marketing efforts that resonate the best with them faster and more accurately. 
&nbsp;

&nbsp;

### 2. Make every penny count
![stack-coins.jpg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/stack_coins_112ad7cafb.jpg)
_(Image by Racool_studio on Freepik)_
 
Running the same campaigns is getting way costlier. The budgets that used to be sufficient to run an effective campaign bring only mediocre results. Many teams are trying to battle the rise in Cost Per Install (CPI) and the drop in Return on Ad Spend (ROAS). As increasing the budget isn’t an option for many companies, teams narrow down the scope and cut more experimental initiatives. 
 
**What to do?**
- Work closely with product teams to produce more marketable games
- Keep a constant pulse on your campaigns and gather a fast understanding of what is working and what is not
- Double down on the best-performing UA channels
 
**How can analytics move the needle?**

UA teams can optimize metrics like ROAS to their full potential by
- Look into campaigns at a more granular level by automating diagnostics analysis.   
- Take a closer look into metrics that looks flat as drilling down can uncover opposing trends/effects and reveal actionable insights to improve metrics
&nbsp;

&nbsp;

### 3. Increase team efficiency and fuel experimentation
![5451475.jpg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/5451475_80c4d86fd3.jpg)
(image by Freepik)

The complex nature of user acquisition in gaming requires teams to analyze vast data sets. With so many factors potentially affecting metrics, teams don’t know where to look and what would move the needle. Trying to do the analysis on dashboards or in Excel means a tremendous amount of manual work. And if your team is smaller than it used to be, capacity is going to be even more limited. Testing millions of hypotheses is humanly impossible. With a large or small team, you will have to sacrifice either comprehensiveness or speed. 
 
**What to do?**
- Invest in augmenting analytics to automate menial repetitive tasks
- Reallocate time spent on manual analysis to high-value tasks and experimentation
- Quickly test new campaigns and iterate
 
**How can analytics move the needle?**
Analytics can focus the teams' attention on areas that matter the most by helping them dig deeper quickly and to understand what's working for which channel/audience and other combinations versus just scratching the surface by testing for the usual suspects. 
&nbsp;
 
 &nbsp;
 
## How do leading teams leverage analytics?

Leading gaming companies observe the behavior of their players and potential users very closely. They collect data points directly from the game as well as from marketing and analytics partners. Pushing the limits of analytics and using all available data, they adopt a more proactive approach. They excel in building better games, engaging players, and maximizing the return of their marketing efforts. In addition, they ensure close collaboration between analytics, marketing, and product teams to share insights seamlessly and make regular improvements. 
&nbsp;

&nbsp;

## In a nutshell
As challenging as it is for User Acquisition teams, it is also a very interesting time. Creativity and experimentation are becoming more and more important. To overcome the market challenges and excel in UA efforts, teams should focus on
- Forming a stronger collaboration between user acquisition, product development, and product management teams and sharing insights and learnings
- Automate repetitive manual analysis as much as possible to enable teams to adopt a proactive approach and spend more time on creative tasks and experimentation
- Measure the return on investment very closely, double down on well-performing activities or improve/pivot quickly on the rest. 


Read Kolibri Games case study

Are your key user acquisition metrics affected by the recent market challenges?

Beat the odds by increasing speed to insights with diagnostic analytics. 

## About the author and Kausa
[Stefan Doerfelt](https://www.linkedin.com/in/doerfelt/) is Co-Founder and Chief Executive Officer of Kausa and [Yagmur Anis](https://www.linkedin.com/in/yagmur-anis-70311120/) is Senior Marketing Manager at Kausa.

Using machine learning, Kausa can analyze all your data, diagnose why metrics are changing, and provide actionable insights to improve business performance. With Kausa teams can save countless hours on data analysis and get ahead of competition by finding opportunities to unlock hidden value and cost savings.

The new rules of user acquisition in gaming

Breaking Six Bad Analytics Habits to Unlock Value

For many companies data is not delivering on its promise (and ROI). Because the data journey is like a marathon and business value and impact start coming in at the last mile.

What is referred to as the last mile of the analytics marathon comprises data analysis, insight communication, and taking action. However, most companies are falling out of the race before tackling these last three steps and, as a result, fail to achieve the expected outcome.
![brent dykes analytics marathon.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/brent_dykes_analytics_marathon_082204b961.png)
(image by [Brent Dykes](https://www.effectivedatastorytelling.com))

Not delivering value can quickly start damaging the data culture and could lead business teams to overlook the efforts spent in the earlier stages of the data marathon. Data practitioners that want to create a strong data culture and have a seat at the decision table should carefully evaluate whether these habits are hindering their teams

## What’s holding data teams back?
While there isn’t a straightforward answer to this question, I have identified six bad analytics habits based on many conversations with data & analytics leaders and practitioners. This list comprises topics that lead to costly decisions and missed opportunities to deliver business impact and are related to the four key elements: people, processes, tools, and culture, which are:


## The six main bad analytics habits

- Treating data/analytics/BI teams like a dashboard factory
- Analyzing only a portion of the available data
- All descriptive without diagnostic analytics
- Waiting too long to act
- Making it impossible to ask follow-up questions
- Failing to standardize and streamline insights stories

The following image summarizes the relation between each bad habit and the corresponding root causes across the four key elements:

 ![Main reasons bad analytics habits.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Main_reasons_bad_analytics_habits_31f1e9ae21.svg)

## 1. Treating data/analytics/BI teams like a dashboard factory
These teams do not have established processes or frameworks to structure workstreams and gather context and information. Hence, data teams receive many ad-hoc requests and are asked to create one dashboard after another without proper context. Often they jump into analyses without asking for clarification or understanding of the ultimate goal (i.e., what is the business trying to achieve). These communication gaps between data and business teams lead to working based on assumptions and data analysts answering the wrong questions.
The culture of asking "why" is also often missing.

**Impact:** Wasted time and delay in critical decisions.

**How to break it:**
- Involve data teams earlier in the process and align on expected business outcomes and what actionable insights look like
- Foster a strong collaboration between data and business teams
- Develop processes to make people reflect on the why.
- Consider embedding your analysts into the business teams.

&nbsp;
## 2. Analyzing only a portion of the available data
Most companies only analyze a small fraction of the data they collect. Teams often just look into the top drivers and the usual suspects (i.e., the usual slice and dice), which results in testing very high-level hypotheses. There are also data silos that prevent teams from leveraging all the available data (e.g., the marketing team only looks into marketing data without combining it with orders or product data).

**Impact:**  Missed insights and opportunities

**How to break it:**
- Go beyond the available slices and dices on dashboards and start analyzing all actionable dimensions in your data
- Go deeper by looking at combinations of multiple factors and leveraging different types of data (e.g., customer, product, and marketing data)

&nbsp;
## 3. All descriptive without diagnostic analytics
Showing what is happening in dashboards is informative but not insightful. Only the “why” behind these changes can drive recommendations and actions. However, teams often look at dashboards in weekly/monthly reviews and comment that metrics went up or down, without clear answers to root causes or actionable insights.

**Impact**: Lacking actionability and missed business opportunities

**How to break it:**
-  Identify where you stand in [the diagnostic analytics maturity](https://www.kausa.ai/blog/diagnostic-analytics-gap) journey and identify the next steps to develop further 

&nbsp;
## 4. Waiting too long to act
Teams spend so much time analyzing that they fail to act on time. As businesses change faster than ever, the speed to actionable insight is getting more relevant. Business stakeholders need quick and actionable insights to make informed decisions before it is too late. In most cases, teams need insights on the same day, and waiting a couple of days or weeks is too long for the business to take action. 

**Impact:** Reactive teams, missed opportunities, undermined data culture

**How to break it:**
- Leverage technology to augment current workflows and accelerate speed to actionable insight
- Spend most of the time on uniquely human value-added-tasks like insights communication and decision-making

&nbsp;
## 5. Getting stuck in a loop of follow-up questions

When diagnostic analytics are performed using dashboards, follow-ups are inevitable. To understand why things are changing and how to act on them more granular insights (e.g., combining 2 or 3 different dimensions) are needed. Manual analysi could require multiple iterations and creates follow-up loops as initial answers are only scratching the surface. However, it is often hard for business leaders to answer their questions without the support of the data team.

**Impact:** Overwhelmed data teams, unanswered questions, analyses that only scratch the surface, and missed opportunities.

**How to break it:**
- Invest in tools & training that empower business leaders to answer their questions
- Consider decision intelligence platforms that enable stakeholders to drill down to the why

&nbsp;
## 6. Failing to standardize and streamline insights stories
Reports and analyses are often distributed across various platforms and lack a consistent structure: Slack messages pointing to dashboards, Excel sheets with ad-hoc computation, PowerPoint slides, Notion pages, tickets, etc,... Even if insights are uncovered, they do not make their way to the decision-makers, or their relation to business priorities is not well-explained.

**Impact:** Difficulty in interpreting results and drawing actionable insights,  missed opportunities.

**How to break it:**
- Establish common formats for reporting. Leverage standard methods like waterfall charts to summarize the key metrics changes.
- Develop and promote best practices for data storytelling

&nbsp;
## Bottom line: focus on the last mile
Starting with the way teams communicate and collaborate and extending to the tools used to uncover insights and the process to distribute them, these bad analytics habits are hindering teams from conquering the last mile of analytics and leading to missed opportunities…

Where to begin?
- Carry out a comprehensive assessment of your current situation to see if you fall into these patterns 
- Focus on what is holding you back the most 
- Pay attention to what is breaking your value chain earlier in the process and causing a domino effect on the next steps

Habits are hard to break. Take your time to assess, prioritize and keep introducing incremental changes. 

&nbsp;

References:
Brent Dykes, [Data Analytics Marathon: Why Your Organization Must Focus On The Finish ](https://www.forbes.com/sites/brentdykes/2022/01/12/data-analytics-marathon-why-your-organization-must-focus-on-the-finish/?sh=2b9a7a4a3c3b)(2022), Forbes.com



Interested in creating more opportunities from your data, 

doing proactive analysis and communicating more insights among your teams?

## About the author and Kausa
João Sousa is the Director of Growth at Kausa. Stay tuned for more posts on how to increase the value of analytics.

[Kausa](https://www.kausa.ai/) accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously.



Why data is not delivering on its ROI for many companies

Breaking six bad analytics habits to unlock value

The diagnostic analytics gap - Why this gap is relevant and how to close it

Showing what's happening in dashboards is informative, not insightful. It's useful to know if a key business metric is going up or down, but it's not actionable. Only the "why" behind these changes can drive recommendations and actions.

However, I often hear how teams look at dashboards in weekly/monthly reviews and comment that metrics went up or down, without clear answers to root causes or actionable insights. This leads to missed opportunities to drive business impact through data and analytics, hindering the agility needed to stay ahead of the competition.

&nbsp;  

## Why is this relevant?
Analytics is still primarily descriptive in most organizations (i.e., WHAT’s happening). 

Few data and business teams have nailed diagnostic analytics (i.e., WHY this happened). 

As external events affect the business and stakeholders take action, metrics change as a reflection. Businesses need to understand what drives KPIs and why they are changing to make better decisions. This can only be achieved by nailing diagnostic analytics: providing comprehensive insights into metrics changes at the speed of business. 

But oftentimes, stakeholders and data practitioners come up with hypotheses on why metrics changed based on their domain expertise, intuition, and experience. These reasons are typically high-level and just scratching the surface by testing 2 or 3 usual suspects, leading to missed opportunities.

&nbsp;  

## The analytics landscape

On a high-level, diagnostic analytics is often overlooked as it lies between data and the business. It’s helpful to compare it against the other types of analytics types. 

![The diagnostic analytics gap.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/The_diagnostic_analytics_gap_b771ec6a4c.svg)

**Descriptive analytics** is self-served through dashboards. Data teams manage governance and infrastructure and enable business teams. There are clear processes, ownership, and tools. 

**Diagnostic analytics** lies between the business and data. It’s nearly impossible to self-serve diagnostic analytics through common dashboards (although many data teams fall into this fallacy). There aren’t clear processes and the data culture varies significantly.

**Predictive and prescriptive analytics** are managed by the data teams with clear data products (e.g., recommendation systems, forecasting models) that are used by the business folks. They are technically complex, but very clear concerning processes, people, and culture.

&nbsp;  

## What’s causing this gap
As data grows in volume and complexity and businesses change faster than ever, the gap between business needs and delivery keeps widening. Most organizations overlook this gap for various reasons. A deep understanding of the gap requires a holistic approach based on the 4 key elements: people, culture, tools, and processes.

Let’s go deeper into each one of the 4 key elements.

![4 dimensions visual.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/4_dimensions_visual_33ad848daf.svg)

## How to close this gap?
Ideally, business and data teams would proactively and regularly analyze metrics changes. Even if the change is minor, there are typically opposing effects. It’s crucial to identify them so that teams can quickly differentiate what’s working and what’s not working and take actions to maximize business impact.

Where to focus depends on the starting point. It requires a comprehensive overview of the 4 elements described before. Based on conversations with dozens of data and business leaders, I often see three scenarios:

The 4 states of diagnostic analytics:
![4 States of diagnostic analytics.svg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/4_States_of_diagnostic_analytics_b03f540850.svg)

**State 1 - _Stuck in the “what”_**

Companies lack the data culture to drill down into the why. Teams monitor metrics on dashboards, but rarely identify data-driven reasons as to why they changed.

**Actions:** Foster data culture and show the value of diagnostic analytics to business stakeholders

**How:**
- Pick 2 or 3 use cases where you see significant potential to deliver business value. 
- Proactively investigate changes in these business metrics and share your findings in a structured and impact-driven manner
- Continue until you achieve a couple of success stories. This should be the base to further expand into additional use cases and start the change management process


**State 2 - _The usual suspects_**

Teams end up only scratching the surface of WHY metrics change and only test a couple of usual suspects regularly.

This is the tricky one because it lacks symptoms. Business stakeholders just test their high-level hypotheses, leading to missed opportunities. Data teams believe that they are doing a great self-service analytics job.


**Actions:** Raise awareness that not drilling down to the why is leading to missed opportunities

**How:** 
- Pick 2 or 3 critical use cases where you know that business teams are only looking at the usual suspects. 
- Proactively drill down to the why and share your findings by comparing comprehensive analysis vs previous analyses that were only scratching the surface
- Build out success stories and expand into more use cases


**State 3 - _Need for speed_**

Business is aware of the value of drilling down to the why but struggles with the speed-comprehensiveness trade-off.

In this case, data culture is already advanced and teams are only missing the last mile of the marathon. Most teams are aware there is business value to be unlocked, but struggle to accelerate diagnostic analytics. Performing comprehensive root-cause analysis with existing workflows is too complex and time-consuming. Teams struggle to deliver insights at the speed of business.


**Action:** Augment current workflows to accelerate speed to insight and eliminate the speed - comprehensiveness trade-off.

**How:**

- Consider augmented analytics platforms to accelerate speed to insight and eliminate the trade-off. These tools leverage ML to automate hypothesis testing and driver scoring. This way teams can get insights at the speed of business and you can scale analytics without throwing more analysts at the problem
- Review current diagnostic analytics procedures and ensure a proper way to conduct these analyses, share insights and organize them
- Increase the collaboration between data analysts and the business for the most important use cases to develop domain expertise

**State 4 - _Full Force_**
Teams proactively diagnose metrics changes for constant improvement. By augmenting traditional workflows teams eliminate the speed-comprehensiveness trade-off, share actionable insights proactively and capture a wide-range of opportunities.

This state can be showcased as a best practice. The teams work collaboratively and proactively, processes are well-defined and clear, and the right set of tools are implemented to drill down to the why and find actionable insights. As a result, the ROI of data analytics is clear and there is strong trust in the data culture. 

&nbsp;  

## Bottom line: the key lies in the why
Showing what’s happening in dashboards is not enough. Drilling down to the why comprehensively and quickly is the only way to deliver actionable insights that increase the value of analytics. This is the way to conquer the last mile of the data journey.












Want to improve issues caused by diagnostic analytics gap, 

start augmenting analytics and foster data culture?

## About the author and Kausa

João Sousa is Director of Growth at Kausa. Stay tuned for more posts on how to nail diagnostic analytics and increase the value of data.

[Kausa](https://www.kausa.ai/) accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously.

Why this gap is relevant and how to close it

The diagnostic analytics gap

Michael Klaput, Ph.D.

Co-founder & CTO

Why analytics needs an impact-driven approach

**The amount of data and its complexity are rising constantly, making it harder and more expensive for businesses to find the right places to create value and drive impact.**

When your marketing team is running thousands of campaigns in different regions, countries, and platforms, there are millions of hypotheses that can be tested and numerous interventions that can be made. Or there are hundreds of different touch-points in a website that might be affecting the way consumers engage or churn and it differs for each consumer segment.

&nbsp;   

## Constantly chasing of why metrics change
This means that your metrics such as return on ad spend, conversion rate, or basket size are changing on a regular basis. But do you know _why they are changing?_ Do you know what to do in order to move the needle in your favor? If you want to create impact, new learnings need to be implemented in your marketing and sales efforts every day. Even for well-performing efforts, failing to track and make incremental changes could result in downturn in the longer run. 

Using traditional business intelligence tools means that you need to know the right question to ask to get started. And it can be both frustrating and time-consuming even to identify the right question. While business teams throw different questions at data teams hoping to identify the reason for metrics changes, data teams are overwhelmed with the requests. And more often than not, by the time the right question and the root causes are found, it is too late to act on it, missing the window to create value.

&nbsp;   

## A new approach is needed…
To get to the why, businesses need speed and scale to be able to transform through data, a way to reduce time to insights from weeks to minutes, and an accurate view of priorities to empower team members to focus on what really matters.

This is where the power of machine learning comes into play:
Every possibility is analyzed within minutes,
actionable insights are put into business teams’ hands in real-time, and
no opportunities are left on the table. 

Given that the abundance of data will only get greater, businesses, especially the ones in fast-growing and fragmented markets, need to navigate this complexity well and fast to stay competitive.

&nbsp;  

## A high level view is informative, but not yet actionable
Let’s take a look into our international, cross-platform campaign example and assume that you are seeing a global uplift of 47% in ROI (Figure 1). That’s an impressive outcome, but to keep up the performance, you need to diagnose what is driving this increase. 

![Fig 1 - Change in global ROI.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_09_21_at_10_59_53_0bc20ab741.png)

When analyzing the ROI of this campaign, what if you could understand how much _every sub-campaign or sub-segment contributes to the ROI_ of your global marketing efforts within seconds? And you could easily view the information _around the platforms, countries and demographics to have a better context?_  What if you could even cross-check the change in ROI with the changes in other _related metrics such as CPI, predicted revenue, or in-app conversions_?

I'm sure you agree that this would be great. But you may be thinking: "_Can't I do this already today in my dashboards?_" In fact, no... Because dashboards are great at showing you changes, but they don't help you see how it all fits together. 

&nbsp;  

## Turning to dashboards for analysis and identiying next course of action could be misleading
Let’s take a look at Figure 2 below, where you can see a simple visualization of ROI change by country. You are most likely going to direct your attention to _Finland, Ireland, and Spain_ as they have the most steep changes. What if I were to tell you that the _main contributors to the Global ROI are actually Croatia, Austria, and Belgium,_ and Finland is actually one of the smallest contributors? 

![Fig 2 - Change in ROI by country.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_09_21_at_10_59_30_613f71e15d.png)

As the Figure 2 didn’t take into account the _size of the marketing spend and revenue change_ into account, it diverted your attention, instead of focusing it on the high-impact countries. Looking at individual changes loses track of how much each piece really moves the needle for your business. When you do take the full picture into account, you can understand by how much each country has been pulling the global ROI up or down. We have summarised the results of such an analysis in Figure 3 below.

![Fig 3 -  Contribution by country.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_09_21_at_10_59_03_3246b95db7.png)

&nbsp;  

## True insights are generated with impact in mind
Working on a small set like this, similar insights can be found through manual slicing and dicing as teams will know the usual suspects to go after, but this becomes impossible as datasets grow, more factors and subfactors come into play and performance comparisons over time are needed. _Depending on gut feeling will only lead to missed opportunities or lost savings._

Just because you have complex data, working with it doesn't have to be complex as well. Decision intelligence platforms powered by machine learning can help simplify and automate your data processes. This means:
- **your business teams** will get informed of potential opportunities within minutes
- **your data teams** are freed from manual slicing and dicing as well as ad-hoc requests and are able to concentrate on strategic priorities and take on a proactive role. 
- **your analysts** are empowered to deliver insights that focus your team’s attention to what really matters

***

## About the authors and Kausa
[Michael Klaput](https://www.linkedin.com/in/michael-klaput-67375385/) is Co-Founder and Chief Technology Officer of Kausa and [Yagmur Anis](https://www.linkedin.com/in/yagmur-anis-70311120/) is Senior Marketing Manager at Kausa.

Kausa accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously.

How to create value and deliver business impact using your data

 How causal inference lifts augmented analytics beyond flatland

**Causal inference techniques along with business analytics approaches unravel what truly changes your KPIs, while predictive modeling falls flat.**

If the world was two-dimensional, life would be very odd indeed. Think about it: The earth would not be a sphere but a circle — just like all other stars and planets in a 2D-universe. Beings would be flattened, too, navigating a plane landscape and existence. For example, to pass somebody on the street, you would have to jump over that person as there would be no depth whatsoever. For the same reason, to just look behind you, you would literally have to turn yourself upside down. Fortunately, this is not the world we live in. But unfortunately, this is the basis on which most enterprises are run today — perhaps even yours. In any technology-driven business, the quality of your Decision-making is inevitably based on the quality of your data insights. However, in too many companies, those ’insights’ are effectively two-dimensional: flat, impractical, and hopelessly inconclusive.

![super mario.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/super_mario_df0af90b8f.png)

Businesses usually measure their performance in terms of a KPI. It has thus become an objective in data analytics to find the best predictive models of future KPI values given historical data. While these models might perform surprisingly well, extracting value out of them is just as hard. Aside from a lack of explainability, this is also because predictive models are unable to capture reality and are limited to low-dimensional explanations. In this article, we will give two arguments why this is the case based on bad scaling and unrealistic assumptions made in most predictive models.

But why bother? It is not the well-performing model that improves business performance. Instead, the only way to improve a business is through decisions, which should ultimately be done by a human. The objective of data analytics in business should be to inform decisions by uncovering insights. Unfortunately, these are hidden in your data like needles in a haystack. [This far from trivial problem motivated a relatively young branch of data analytics has been coined augmented analytics and has been pushed in a recent Gartner report.](https://www.gartner.com/en/doc/441420-how-augmented-analytics-will-transform-your-organization)

We want to challenge the perception that predictive models should be the default option used to inform business decisions. They introduce a costly detour in the search for insights, and might even render it practically infeasible. We will highlight in a simple problem that predictive modeling can offer only little aside from a massive overhead. Instead, we will try to mimic how a business analyst would operate. This will naturally bring us to methods from causal inference.

&nbsp;  

## The Impactful Change
 
We will consider the problem of diagnosing errors in regression models in big data scenarios. Most of the readers should have encountered the following scenario:
![KPI actual vs prediction.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/KPI_actual_vs_prediction_cfedbccfc5.png)

Clearly, there has been an impactful change in the KPI that seems to persist for the time being. On the technical side, a reasonable reaction would be to retrain your predictive model on the data after the change point. Do you agree? If so, maybe keep this in mind.
![KPI actual vs prediction two models.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/KPI_actual_vs_prediction_two_models_964d6e8388.png)
The good news is that your model seems to be accurate. The bad news is that your manager will inevitably ask what happened to the KPI. But do not be afraid. This is an ideal situation for proving your value to the company. Can you identify the reason behind this change? Can you uncover insights that will inform the correct decision?

&nbsp;  

## The Quest of Why

Let's say your company’s dataset looks something like this:
![quest of why.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/quest_of_why_92e996bb8f.png)

Let us further assume that you are dealing with the simplest case: The KPI values of data points before and after the jump are perfectly fit using a linear regression model. Here, you are dealing with categorical data which you need to handle appropriately to use in the regression. A standard approach for this is one-hot encoding of categorical values: for each category value, you introduce a feature that can be either true or false. For example, in the dataset above you would define the feature that can be either true or false. For example, in the dataset above you would define the feature customer_country == Germany. To finally enable feature selection, it is necessary to use a form of regularisation. Here, you will use lasso regularisation (with ten-fold cross-validation).

After training two lasso regularised linear regression models, one before and one after the jump,  you can look at a ranked list of feature weight differences between these.

![sample data.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/sample_data_05d85a45c2.png)
![feature weights.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/feature_weights_19406e086f.png)
It looks like subgroups such as android customers or customers older than 46+ performed differently before and after the jump. Great, looks like you found the reasons for the KPI jump … or did you?

&nbsp;  

## The Curse of Dimensionality


In fact, this is a more non-trivial situation than we have appreciated so far. Imagine presenting this to the KPI owner. They will be very happy that you have delivered them reasons for the KPI change and they will now be wondering about what to do based on this information. It will automatically lead them to questions like the following: _“Are the actual drivers for the KPI change all android-tv customers, all customers older 46, and all customers who made a purchase before? Maybe it could be the repeat customers older than 46 and the android-tv customers … or the android-tv customers who purchased something before? Worse, are there maybe other combinations of features which you have missed?”_

Therefore, to be able to more confidently answer such questions, you would have to repeat your regression analysis with more complex one-hot encoded features … now representing finer subgroups than before. Thereby, you are searching among deeper subgroups of the dataset, see Figure 6, with new features like customer_age == 46+ AND first_order_made == yes, customer_age == 18–21 AND first_order_made == no.

Again these subgroups enter via one-hot encoding. This is obviously problematic, as you are now falling victim to the curse of dimensionality. [It is the era of big data, and you just increased your number of features by a factorial amount.](https://medium.com/swlh/stop-one-hot-encoding-your-categorical-features-avoid-curse-of-dimensionality-16743c32cea4https://medium.com/swlh/stop-one-hot-encoding-your-categorical-features-avoid-curse-of-dimensionality-16743c32cea4) A piece of code that can be used to generate these refined subgroups is

![sample data 2.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/sample_data_2_2860135dbd.png)
![subgroup depth.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/subgroup_depth_d4ffb4ac56.png)

Remember that linear regression is based on an inversion of a covariance matrix among all features - which scales O(d³), with d being the number of features, i.e. in our case the number of possible subgroups. This introduces significant opportunity cost in comparison to non-predictive feature selection methods — as will be discussed later.

![sample data 3.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/sample_data_3_4302d6b8b9.png)
![weights of intersections.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/weights_of_intersections_8ed815288e.png)

After some time your computation finishes. While your earlier computation took just 0.1 seconds, searching for third-order features already took over a minute. But it seems to be worth it. You find that the number of groups driving the KPI change was actually one, see Figure 7. Presenting this insight to your manager, he could quickly point to an update that directly affected the subgroup you reported.

&nbsp;  

## By refining the subgroup, you could render it actionable.

While your regression approach finally worked out - it took extremely long to compute resulting in opportunity cost to your company. In a realistic scenario of big data, your approach would have failed horribly. Additionally, original sets containing only shallow subgroups painted an incorrect picture. Only after refining sets and enormous computational effort, you could pinpoint the actual subgroup that drove the jump in the KPI.

This begs a couple of questions:
- Do you actually need to learn a predictive model to answer why the jump happened?
- How can you reduce opportunity costs?
- How can you find subgroups at the appropriate level of granularity?
- Is it economic to retrain models at every jump for this piece of information?

While answering all these questions is out-of-scope for this post, we will offer a new point-of-view that can help to resolve these issues. For this, we will develop an approach to feature selection that improves on linear regression.  Augmented analytics depends on it.

&nbsp;  

## Learning from Business Analysts and Causal Inference

Let’s take a step back … what happened here? You started off with a predictive model, and you saw that it could neither predict nor explain the observed jump in the KPI. Why is that? Because predictive models are unable to capture reality. [They assume that all data is independently and identically distributed.](http://arxiv.org/abs/1911.10500) However, in real-life applications, this is often incorrect, as this example shows. Data before and after the jump was generated under different conditions. You even intuitively made use of this fact, when you used two separate predictive models which (after some tricks) helped us to uncover the reason for that jump.

As you had to give up on prediction and ultimately did not predict anything, what did the prediction models actually do for you? If you think about it, the key is that you are not interested in predicting the KPI as a function of all possible subgroups - you are interested in subgroups affecting the KPI! Thus, to search for insights at deeper levels you have to get away from predictive modeling. This is where the data scientist can learn from the business analyst.

A business analyst searches for insights through dashboards containing meaningful summaries of data. Instead of correlating all features together, as in the regression approach above, the business analyst will try to pinpoint what changes happened in the data based on summaries (like means, histograms, or metrics) by iteratively filtering the data for different conditions. Most importantly, the business analyst will never have to look at all features at once. How do you teach a machine to do that? How can you learn from a business analyst?

Let us formalize the above in mathematical notation. Let X be a subgroup, e.g. X = customer_age == 46+ AND first_order_made == yes and

![formula 1.gif](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/formula_1_3a1bbd2e7b.gif)

some summary of the KPI distributions before and after the jump in the KPI. Then, you introduce conditional summaries

![formula 2.gif](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/formula_2_ac111bdaf3.gif)

where you compute summaries of subsets of KPI values for which X is true. All that our method needs to do now, is to compute conditional summaries for each subgroup and rank them. I want to stress, that in practice these abstract summaries can be objects as means, histograms e.t.c.


[The procedure detailed above is actually a common technique from causal inference.](http://proceedings.mlr.press/v67/gutierrez17a/gutierrez17a.pdf) You thereby implicitly changed our point of view. Now, you consider the mysterious jump in the KPI as an intervention, that is now assumed to have happened due to external or internal treatments. An example for an external treatment might be the holiday season, an internal treatment might be an ad campaign, a change in pricing, or, as in our case a software update. You are thus explicitly lifting the wrong assumption that all data is independently and identically distributed. You are now searching for subgroups that are causal to the change in KPI.

&nbsp;  

## The Quest of Why - Revisited

Now that you have a model of how a business analyst operates, let us proceed with the actual implementation. [For now, you will use a standard summary used in causal inference called Conditional Average Treatment Effect (CATE)](https://towardsdatascience.com/be-careful-when-interpreting-predictive-models-in-search-of-causal-insights-e68626e664b6), for which our summary becomes

![formula 3.gif](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/formula_3_3f5fd2a6f1.gif)

The CATE corresponds to the change in the mean of the KPI, conditioned on that subgroup X is true. Ranking by magnitude then gives us the correct subgroup as a result. To detect multiple subgroups, we repeat this procedure after removing the best performing subgroup after each iteration:

![sample data 4.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/sample_data_4_7e8c0b2baa.png)
![CATE scores.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/CATE_scores_8e4658127b.png)
This takes a fraction of the cost of our predictive model. The computation for first-order features took just 0.02 seconds, searching for third-order features took less than a  second.

Let us take a step back and compare this approach with the earlier one based on regression and what their respective objectives are. Feature selection via regression answers the question: ” Which subgroups best predict your KPI?”. While taking the view of causal inference answers the question: “Which subgroups had the biggest causal effect on the KPI?”.  Comparing run-times of a naive implementation of CATE with the optimised sklearn implementation of linear regression in Figure 9, we find that they lie order of magnitudes apart. This makes it clear that these questions, while superficially similar, have fundamental differences.

![Long runtime over subgroup depth.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Long_runtime_over_subgroup_depth_a662ac7c54.png)

&nbsp;  
&nbsp;  

## Conclusion

Predictive models have strong shortcomings as means to understand KPI changes, especially in multi-dimensional contexts. These models fundamentally answer the wrong questions under the wrong assumptions. Instead, business analytics focuses on why did something happen rather than what will happen. Having their mind free of the auxiliary task of predicting future KPI values, analytics finds reasons in the data to understand why the KPI changed, trying to find the answers for the right question.

Be wary next time you want to explain anything. Firstly, you should ask the right question. In addition, multi-dimensional contexts require a scalable technique based on causal inference and business analytics methods. This is our mission at Kausa: scale business analytics logic and couple it with causal inference to provide the right answers to KPI changes.

PS: [Code and data to reproduce results from this article are available in our Github repository.](https://github.com/kausa-ai/blog/tree/master/how_causal_inference_lifts_augmented_analytics_beyond_flatland)

&nbsp;  

## About the authors and Kausa

Michael Klaput is Co-Founder and Chief Technology Officer and Dominik Linzner is Causal Inference Engineer at Kausa.

_Kausa accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously._






























Unravel what truly changes in your KPIs

How causal inference lifts augmented analytics beyond flatland

Leveraging Data Analytics to create business value

**This article dives into the challenges faced by industry experts in integrating data and analytics into their work streams, and how Kausa fits into the equation to mitigate the resulting effects of the aforementioned challenges and empower data-driven decision making.**

With the world slowly heading towards a digital revolution, the importance of data and analytics is also consequently becoming increasingly rampant. Data-driven is no longer the goal, it is the requirement. However, the desire for this transition alone does not make the integration any easier. Despite a universal agreement on the power data-backed decisions hold, many companies are still struggling to make that a reality and reap the benefits.

According to a study by [New Vantage Partners](https://www.newvantage.com/wp-content/uploads/2018/12/Big-Data-Executive-Survey-2019-Findings-Updated-010219-1.pdf), only 31% of executives feel they have been able to build a “data-driven culture.” And according to [Gartner](https://www.gartner.com/en/newsroom/press-releases/2018-12-06-gartner-data-shows-87-percent-of-organizations-have-low-bi-and-analytics-maturity), 87% of companies have low BI and analytics maturity. Experts in the industry have reported multiple challenges that they have faced in trying to embrace a data-driven culture, which is summarised below in 3 broad categories:

&nbsp;  

## 1) Scaling Data Analytics

As organisations grow, so does the amount of data they collect, which makes analytics increasingly difficult to navigate. It is close to impossible to increase the resources allocated to data (budget, team, hardware, etc.) at the same rate as the volume of data increases. A major problem area around scaling the use of Data & Analytics is also underestimating the security implications. Substantial amounts of diverse data are usually in one place, which creates potentially new vulnerabilities and represents new types of business risks to mitigate.

Tools like AI and ML can do remarkable things with a lot of data, but to scale analytics the two key questions are "do we have the right data?" and "are we exploiting the data in the best way possible?" Answering yes to both of these is indeed a challenge most organisations are working to resolve. The real problem occurs when you try to combine data across various sources. For instance, combining sales data with marketing, or supply chain data with sales; that’s when the cat comes out of the bag. Though these IT systems for Sales, Marketing etc., in isolation work like a charm, when you try to build pipelines to combine them Data Quality/Integrity issues hamper your progress to scale analytics.

&nbsp;  

## 2) Developing data and analytics-driven culture, literacy, and employee behaviour
 
What does it really mean to have a truly "data-centric culture?" This means that every single person - irrespective of expertise, tenure, or department - should be enabled to make better decisions based on data. This is something that is seldom found on the company level, although everyone seems to be wanting this. So the question that begs: what is really stopping them from achieving that data-driven culture status? This can be broadly categorised as follows:

### a) Inaccessible data

In the quest of becoming data-driven it is not only important for each individual to have the necessary toolkit to deal with data, but also what data they have available and how that data can influence decision making. The quality of data most companies possess is rarely up to the mark that it can be used in ways desired. As more decisions are augmented by data, organisations must consider where these decisions are taking place and reduce the frustrating amount of time spent moving between these applications.

### b) Apprehension Over Changing Roles & Skills

[According to a study conducted by INC](https://www.inc.com/scott-mautz/science-says-this-is-why-you-fear-change-and-what-to-do-about-it.html), neuroscience has shown that uncertainty feels similar to failure in our brains. That’s why so many people would rather avoid organisational change because of how uncomfortable the process of learning something unfamiliar can be. As the importance of analytics begins to seep through the foundation of an organisation, some teams may find themselves caught off guard, scrambling to understand how their skills transfer, if they do at all. Executives must not be hyper-focused on creating a data-centric culture to the extent that they do not associate any human element with it. They must think beyond policies, systems, and organisational charts to consider the psychological elements behind developing a data-first atmosphere.

In addition, to break long-held habits and encourage the use of data beyond a core group of analysts, organisations should invest in educating their entire workforce, irrespective of the department they work in, to become more data-literate and explore ways to not only create a streamlined experience for newer users but also find cross-functional common ground between departments, supplemented by data.

&nbsp;  

## 3) Monetising data and analytics, and treating them as assets to maximise economic value

It is never enough to simply have data. The value of data comes from the insights it generates, the processes it optimises, and its ability to enable better decision-making. The reality is, despite data and analytics hype and expectations, most organisations are not successfully monetising their data. It’s not uncommon for organisations to keep investing in systems even when they don’t produce the benefits they had promised. The real challenge comes in generating value in the form of increased sales, revenue, installs etc from the aforementioned data. Many organisations look into making significant investments in data assets, but without the appropriate strategy, they run the risk of putting the cart before the horse and making poor and regressive investment decisions.

&nbsp;  

## Kausa & Data Analytics

Having the right set of tools can, at the very least, facilitate mitigating the effects of the aforementioned challenges. Kausa is key on your road to become a data-driven company by enabling anyone to run an analysis with a few mouse clicks, and no technical expertise required. Ad-hoc requests can easily be answered by adding one filter or just drilling down on a specific subgroup.

![Filter and group comparison.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Screenshot_2022_08_24_at_17_11_02_b278659d89.png)

Kausa's tool will do all the manual work for you, without missing any corners or making errors humans may be prone to do. You can easily delegate the heavy lifting to Kausa, which hypothesis-tests all factors to find the reasons why your metrics change. So you won't have to worry about how to make the best use of the data, and just focus on the insights Kausa has already generated for you. These all-encompassing insights then translate into what you and your organisation essentially work towards - maximised economic value in the form of increased installs, revenue, or margins.

Whether you have employees with no technical knowledge feeling apprehensive towards a data transition, or experienced analysts looking for more time on their plate to be able to focus on more strategic projects, Kausa will have you covered. Its collaborative features allow users to easily share one analysis link on Slack or simply tag a colleague and comment. This way multiple teams can provide their inputs and business expertise on a central insights platform, fostering a more modern data culture and data-driven decision making.

&nbsp;  

## Conclusion

While Kausa does not promise to transform the industry (yet) and solve every challenge your organisation faces, it can certainly facilitate in creating the data-driven culture you have been working towards creating. So you won't have to worry about your employees feeling alienated and not knowing the data's applicability across different departments.

**Interested in trying out Kausa?** [Get in touch](https://meetings.hubspot.com/kausa-because-you-know-better/introduction) with us.

_Kausa accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously._

### About the authors and Kausa

_Usman Amjad_ is the Business Development Executive and _João Sousa_ is the Customer Success Manager at Kausa.

_Kausa accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously._

 




How to overcome UA challenges with data-driven decision making

Leveraging data analytics to create business value

User Acquisition in Gaming Industry

**Optimizing User Acquisition(UA) efforts is a key challenge for gaming companies. Given the fast-changing and complex landscape, dedicated UA teams focus on crunching data and gathering insights to take critical business decisions. This article explores how Kausa can help you maximize user acquisition and LTV.** 

User acquisition (UA) is the continuous process of acquiring new users to download your game or app. It can be done through various methods, either paid or organically, aiming at getting more installs and attract profitable users.

On one side you have the cost per install (CPI) for each user installed through your campaigns. On the other side, you have the predicted lifetime value of acquired users (LTV), which is a combination of retention (how long people stay in your app) and monetization (how much they spend in your app, plus advertising). A positive ROI is achieved by having a lifetime value superior to the cost per install. Depending on the segment of mobile gaming, the lifespan of a user ranges from a few days to several months or even years.

&nbsp;  
&nbsp;  

# UA challenges

## Fast-changing landscape

One of the most critical challenges of user acquisition is not only to make a successful game but keep it running with a high number of DAU (daily active users) and revenue over an extended time period.

We often hear from companies hitting a plateau or decline with user acquisition: "It worked great in the beginning. We grew very quickly and our user acquisition was really profitable. Now we've hit a wall and profitability is going down". Once a game is launched, the signals are usually very obvious in the beginning. However, as it grows, it becomes harder and harder to understand what may move the needle and find the right insights.

One channel that worked very well in the past (i.e., very profitable) may not be the right one anymore. Thus UA requires regular assessment and analytics to optimize ROI.

&nbsp;  

## Multi-dimensional complex analytics challenge

CPI changes based on different budget levels can be represented only if you look at it on a specific level of granularity, as overall CPI on a wide global UA scale depends on many factors, such as:
- Country share: which countries do you spend the most and get the most installs
- Platform share: Android vs iOS
- Channels used: Facebook, Instagram, Google Ads, Apple Search Ads, Video Networks, etc
- Optimization methods: optimizing towards the cheapest possible installs (app install optimization), chasing payers (app event optimization), or higher-value users (value optimization)
- Scale: how much are you pushing with a daily scale of budget/installs on the country/platform/channel level
- Seasonality: CPI varies over the year

UA teams analyze the available data to identify critical segments, testing every potentially relevant factor and combinations of factors. Variables such as country, campaign, device, channel, media source, and game type may play a critical role, just to mention a few. Therefore analysts spend a significant amount of time drilling down on data using Excel sheets and/or BI tools (e.g., Looker), which have significant shortcomings to analyze multi-dimensional data. You can read more into how Kausa is closing the gap between dashboards and data-driven decision making 🌍.

![UA in gaming- a complex problem.jpeg](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/UA_in_gaming_a_complex_problem_c953d8bff6.jpeg)

&nbsp;  

## Agile trial and error over time

UA optimization involves fast experimentation using regular interventions (e.g., launch in a specific country to assess the potential of a new game). These interventions can be segmented into acquisition efforts (e.g., marketing artifacts) and in-game interventions. It requires agile trial and error to smoke-test game versions, various visuals and ultimately increase revenue. Given the sheer frequency of interventions, it's hard to isolate effects and assess the impact of a specific intervention. 

Moreover, the time lag between one intervention and the corresponding effect makes it harder to quantify its impact. Especially as there aren't appropriate tools for unsynchronized time-series analysis with noisy data.

&nbsp;  

## Competitive interaction and other key factors

While continuous data analysis to derive actions is a key part of the UA process, there are other relevant factors and workstreams. Firstly, it's key to track competition and industry trends, as well as gather insights from key partners like Facebook, Google, peers, etc. This way you can take these factors into account when optimizing your budget allocation. Secondly, we know that you also spend a significant amount of time briefing and partnering with creative teams to design the next iteration of ads. Last but not least, activities like brand building are key for long-term growth.

&nbsp;  
&nbsp;  
&nbsp;  

# How Kausa can help you maximize player acquisition and LTV

Rapidly diagnose the most effective player acquisition, engagement, and revenue strategies with Kausa. The challenge to drive player acquisition and monetization is to identify what works and what doesn't early enough to act upon and drive profitability growth. Kausa provides automatic insights at every stage of the funnel. 

Kausa is a data exploration tool to understand KPI changes in a fast and comprehensive way. Its sweet spot is around fast-changing KPIs in multi-dimensional data challenges, like user acquisition. Kausa's AI-driven technology hypothesis tests all potential factors and combinations of factors in the data to identify the causes behind every change in a matter of seconds.

Kausa can play a key role in your UA analytics efforts by accelerating data exploration, delivering automatic insights, and enabling comprehensive root-cause analysis.

&nbsp;  

## Automatic insights

Kausa delivers automated insights, empowering UA teams to make day-to-day decisions by having these insights at their fingertips. This shift from reactive to proactive analytics unlocks significant business value as actions can be taken before it's too late. This is especially critical given that UA is a very fast-changing landscape. In other words, what works this week may not work so well next week, so you need to constantly optimize campaigns.

![Product - impact focus.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Product_impact_focus_3d938b052a.png)

&nbsp;  

## Comprehensive root-cause analysis

Kausa hypothesis tests all factors and combinations of factors in your data so that you don't miss out on any insights. It also conducts automatic feature engineering to make sure sensible features are tested. Thus it delivers granular insights (e.g. combination of country, channel, and platform) to really find out what's impacting your KPIs, like CPI. You can read more about the technology behind Kausa in this article 🌍 by our CTO Michael Klaput.

![Product Hero1.png](https://kausa-headless-cms-prod.s3.eu-central-1.amazonaws.com/Product_Hero1_f950052d93.png)

&nbsp;  

## Accelerate data exploration

Kausa delivers insights in seconds after a few mouse clicks, freeing up analysts to focus on applying their domain expertise and deriving actions, instead of spending hours per day on manual slicing and dicing on BI tools and running tests on spreadsheets. This way you can focus on deriving recommendations and actions, ultimately optimizing campaigns based on the insights delivered to you by Kausa.

&nbsp;  
&nbsp;  
&nbsp;  

# Conclusion

In a nutshell, UA represents a significant analytics challenge given the fast-changing landscape, multi-dimensional complex data, and relevance of competitive interaction and other factors. That's why most gaming companies have dedicated UA analytics teams, whose critical mission is to continuously monitor and optimize UA budget ROI.


Kausa can accelerate data exploration, delivering insights in seconds and at the right granularity to derive action (e.g., a combination of country, channel, and platform). This way you know where to focus your domain expertise, instead of spending hours on data crunching and manual slicing and dicing. Given that you get automated insights in seconds, you'll never feel like you're acting too late. In addition, by hypothesis-testing all potential factors and combinations of factors in your data, you'll feel confident to derive actions and recommendations without being afraid of having missed out on a key insight.

### About the authors and Kausa

Stefan Dörfelt is Co-Founder and CEO and João Sousa is Customer Success Manager at Kausa.

Kausa accelerates data exploration, delivering actionable insights in seconds by testing all hypotheses comprehensively and continuously.







