To help us understand the convergence of analytics and business intelligence (BI) — which, let’s be clear, has been happening for quite some time — we can look at it through the lens of product development, notably based on the Kano Model. Essentially, the model states that each product has different levels of functionalities — basic features and requirements a product needs to be competitive, “excitement” features that are common, satisfy various pain points, and create a positive customer response, and the the “performance,” features, meaning the above-and-beyond, push-the-boundaries features that help customers decide between your product and someone else’s.
This same mindset can be applied across the shifts happening with business intelligence. For a period of time, something is a hot feature, then it moves a step down to being an exciter, then, at a certain point, it becomes a basic feature.
BI With Traditional Reports
BI is not a new thing, it has been around for several decades. In a relatively basic manner, teams have been able to use data in their original systems and repetitive features to generate and deliver certain reports. That can be, for example, the summary of business transactions to provide an annual report. These reports are based on everything that has happened based on the transactions and data in said systems.
In order to provide real value to business leaders, more reports are generated on a more frequent basis, (e.g., on a monthly or quarterly basis). Teams would click a button, a report was generated with tons of files, analysts put the data together in unwieldy spreadsheets, did an analysis, compiled findings in a deck, and presented to the board or key stakeholders. So, a lot of manual repetitive work that previously kept entire departments busy.
From Automation to Automated BI With Dashboards
If we start with automation, enterprises have been performing and automating repetitive tasks for a long time now. Additionally, in the last five to 10 years, teams have moved toward integrating these aforementioned siloed original systems and data together, in the move toward data lakes and warehouses. The goal is, of course, to provide more valuable insights from these consolidated reports — which hits the very core of “business intelligence.”
Slowly, this trend moved into getting an overview of what’s going on with the business via up-to-date, automated dashboards. This only makes sense to do in an automated fashion or teams run the risk of doing this quite repetitive task manually with rather low operational efficiency. This explains why almost every data warehouse has (automated) extract-transform-load (ETL) capabilities of some sort to pull the data together — and this might just be a simple copy job!
More often than not, automation and BI go hand in hand. Teams want to have a report produced every day or every week, which has become the standard (versus just a few times a year, for example). BI can’t be done without the automation behind it and both are core to business operations. But BI overall looks to the past at things that have already happened — forecasting and more advanced analytics are still things that some experts perform by hand. And that’s where AI and machine learning (ML) come into play.
Supercharging Automated BI With AI
Now, data can be pulled together and combined automatically in a so-called ETL job, made available in a consolidated form, and teams can leverage more data, produce more models, and answer business questions that were previously not possible. For the purposes of simplified understanding, we can call this BI 2.0. So, this new and improved BI (a core function of the business that highlights what’s going on and shapes the decisions teams will make) is morphed with the “future looking” features of AI and ML — we extend the purely backwards looking reports with credible forecasts on our KPIs.
To arrive at actions based on forecasts instead of historical data, teams need to look at a business question in the inverse. Instead of looking at a specific KPI to reach a certain revenue threshold and making a forecast for revenue in six months, it’s much more interesting to say, “I would like to hit X revenue threshold, what actions do I need to take to get there?” In order to answer the question, we would need to implement an automated workflow of available data, assess the BI and individual data sources, and infuse AI to help inform the steps for how to reach that certain goal.
Fusing All 3 Together
It’s like building blocks — you can’t do AI without the BI being done to a certain point. Which is where the automation comes in. You don’t want a workforce of data analysts copying data together manually, that’s inefficient and wastes valuable time. The same is true for providing advanced analytics like forecasts in a manual fashion. So, morphing automation, BI, and AI together as one business function — while organic — is pivotal because, after all, AI is only useful if it’s being used to solve a business objective.
At the end of the day, it makes sense that these three building blocks are almost converging — we don’t want to have an ivory tower with siloed tools, so by bringing them together with a purpose (helping teams understand the pulse of the business, how to develop/evolve it in the future, etc.), it enables them to steer the business in the right direction. As we move into 2022, we believe the confluence of analytics and BI with data science and ML will become one practice for the organization, which is a positive change because — in practice — it reduces silos and the separation between people/teams working on BI and AI initiatives, which helps improve cross-team project visibility and collaboration.