The AI For BI Value Framework 2.0

Read time: 7 minutes

Hi there!

Today, I'm going to walk you through my revamped "AI for BI Value Framework 2.0", which is all about enhancing your business analytics using AI right from the start.

Many people believe that mastering BI should come before delving into AI. However, I disagree.

I view AI as a catalyst for modern BI systems, enabling them to unleash their full potential throughout the entire analytical process.

Now, let's dive in and explore how this synergy works!

AI is changing the BI landscape forever

To clear things up front: Traditional BI systems will still be widely used for basic business reporting and insights, and they're not going away anytime soon.

But the way most organizations approach their current BI platform is to meet analytical needs, not to achieve analytical excellence.

Yet analytical excellence is exactly what's needed to stay ahead in today’s competitive business landscape. In fact, when you look at the largest companies by market capitalization, nearly all of them call analytics as one of their core competencies.

There are several key challenges to attaining analytical excellence:

• The ever-increasing volume of data

• The escalating complexity of data

• The demand for rapid insights derived from data

• Limited data literacy and data culture within organizations

This is where AI comes into play.

The AI For BI Value Framework 2.0

The AI for BI Value Framework maps the impact (value) that AI can have on analytics (BI) across three broad themes:

• Making unstructured data available for analysis

• Making it easier to get insights

• Making better predictions.

Here's what the framework looks like:

It's important to note that each of these stages requires a human in the loop to control the entire process.

That's why I like to think about "augmenting" analytics with AI rather than "automating" analytics with AI.

Different areas of value gain require different levels of data literacy.

For example, users with less analytical skills may benefit from querying their data using natural language, but they may not benefit from tools like Auto ML. More advanced users, however, will see Auto ML as a lever to achieve even better results given their advanced data analysis background.

Let's explore these value areas step by step.

Value Area 1: Making unstructured data available for analysis

Traditionally, business intelligence systems work with structured data from relational databases. However, with the rise of digitization, the amount of unstructured data in text, image, or audio files has increased significantly.

Welcome to the first part of the AI for BI Value Framework:

AI can help analyze unstructured data at scale and enhance BI capabilities by extracting structured data from unstructured sources.

At this stage, you can take advantage of four of the five AI archetypes:

• NLP will help you understand unstructured text (e.g. customer reviews),

• Audio AI services will help you to make sense of speech data at scale (e.g. call center analytics),

• Computer vision lets you turn images and documents into structured data, and finally

• Generative AI modelssuch as Large Language Models (LLMs) are headed to combine these different types.

Here are two examples of how AI can unlock new data sources in this value area:

Example 1: Computer Vision

AI can use computer vision to analyze images, identify patterns, and convert this information into structured data for further analysis by BI systems. For example, computer vision can be used to monitor inventory levels in warehouses or detect manufacturing defects in products.

• Further reading: Counting cars for traffic management

Example 2: Natural Language Processing

AI can use natural language processing techniques to extract structured data from raw text files or PDF documents, making it compatible with BI systems for better prediction and understanding. A practical example is analyzing customer reviews to identify common themes and sentiments to improve products or services.

• Further reading: Analyzing customer reviews

By incorporating these additional insights from previously untapped unstructured data sources into BI analysis, organizations can gain richer insights and make more informed decisions.

Value area 2: Making it easier to get insights

Traditional self-service BI systems often still have a lot of technical friction due to their complex data models, making it difficult for non-technical users to gain insights.

This is where the second part of the AI For BI Value Framework sets in:

At this stage, AI can help you to reduce the friction and create a more intuitive user interface.

The following AI archetypes play a role here:

• NLP helps you to build powerful text interfaces that allow users to ask questions about the data and summarize analytical findings

• Audio builds on top of NLP, allowing voice interfaces for dashboards

• Auto ML helps to accelerate the discovery of insights by automatically analyzing patterns at scale

• Generative AI allows you to build even more powerful analytics chatbots or generate individual charts just by asking for them.

Here are three concrete examples:

Example 1: Natural Language Q&A in Reports

Advances in natural language processing (NLP) allow users to ask questions in plain language, making it easier to interact with BI systems. For example, Power BI's Q&A tool allows users to ask questions about their data set, such as

Example 2: Automated Pattern Detection

AI-powered algorithms can quickly analyze data and provide insights by identifying patterns, correlations, or unusual observations that humans might miss. Power BI provides visualizations such as the Key Influencer and Decomposition Tree visual to help users explore data using ML-based techniques.

Example 3: Explain Analytical Results

AI can create summaries based on data, making it easier to communicate key findings and reducing the risk of misinterpretation. For example, Microsoft Power BI can automatically generate chart labels, such as "Sales $ for Texas increased for the last 5 years on record and experienced the longest period of growth in Sales between 2010 and 2014." These auto-generated snippets save time and make reports more accessible.

As you can see, there are several ways that AI can help reduce technical friction in traditional self-service BI systems, making them more intuitive and easier to use, especially for less-technical audiences.

Value Area 3: Making better predictions

Traditional BI systems have struggled with forecasting, often resulting in crude heuristics with limited actionable insights.

That’s the final impact area of the AI for BI value framework:

At this stage, there are two major AI capabilities at work:

• Auto ML helps to improve predictions from tabular data or complex time series patterns

• Generative AI supports this process by letting users make inference from the data itself (e.g. “given the following data, which 100 customers are most likely to churn?”

Here are two examples:

Example 1: Micro-Level Predictions with Auto ML

For example, instead of just predicting the total amount of recurring customers a business will have in the next month based on aggregated retention metrics, AI can calculate a churn probability for each individual customer, resulting in both optimized forecasting and also improved actionability for targeted retention measures.

• Further reading: Predicting customer churn

Example 2: Capture Complex Times Series Patterns With Auto ML

Specialized time series prediction algorithms detect patterns in larger data sets, leading to more accurate predictions over time. AI capabilities can be deployed as part of existing BI software or independently at the database level.

• Further reading: Demand forecasting with AutoML

Techniques like auto-ML and generative AI let organizations leverage advanced analytics capabilities with AI without employing a large number of data scientists or ML practitioners, as long as users are able to understand the underlying data and analytical concepts.

Conclusion

AI is transforming traditional BI by accelerating data analysis, democratizing access to insights, and leveraging machine learning methods to prepare data for analysis. This transformation improves decision making, usability, forecasting, and data accessibility.

The ideal AI-driven BI solution combines both automated and human decision-making. Keep this in mind as you develop your own AI-integrated BI use cases.

That concludes today's edition. I hope you enjoyed reading this article.

If you found this newsletter useful, please don't hesitate to share it with your colleagues or send me your feedback.

I always look forward to hearing from you, and I promise I'll respond to every message!

See you next Friday,

Tobias

Want to learn more? Here are 3 ways I could help:

1. Read my book: Improve your AI/ML skills and apply them to real-world use cases with AI-Powered Business Intelligence (O'Reilly).

2. Book a meeting: Let's get to know each other in a coffee chat.

3. Follow me: I regularly share free content on LinkedIn and Twitter.