I see BI (business intelligence) as applying and deploying your business logic. You want insight into your business. Using data, you can recognize what's working and what's not. You want to be able to copy and paste that in an easy way for your business. It is estimated that 90% of all data in the world was generated only in the last two years. Bizarre that those volumes are increasing so much! Of course, you want to take advantage of that in a smart way.
So one is BI, repeating the business logic and making it insightful. So BI can also be automating certain processes. That can be with a digital robot that categorizes your emails or keeps your Sharepoint tidy. It can also be mathematical analysis to get those insights. Perhaps you already use dashboards, so you always have insight into how your business is doing. These dashboards provide insight into your capacity or quality, for example. This allows you to define very targeted areas for improvement and actions. Data-driven work at it's finest. Okay, great analyses and dashboards. Do I need them to be able to predict?
In my opinion, you don't necessarily need BI to start forecasting. But, you do need to know your numbers. Prediction, or the use of AI, can also predict a certain action. A certain outcome. Think about whether a particular machine will break down in 6 months. Or what the amount of your sales will be. Or which drug is best for that patient. But ask yourself: is all this really AI?
Did you know there is a distinction between AI and data science? AI is data science, but not every data science solution is AI. AI means intelligence . An AI model learns from its own process. It improves as the model is used more often. More data and more frequent use means the model becomes more intelligent. Think of a self-driving car. The more data there is (namely, driver behavior), the better the car learns to "predict" that behavior. Because that's basically what that car's AI does: predict human driving behavior to turn that outcome into automated actions.
There are different techniques within AI and data science that you can employ, depending on the type of issue and the type of data. If you want to predict a value, you need to know what kind of data you have:
- Nominal: categories without order (e.g., product types or gender)
- Ordinal: categories with order (e.g., customer satisfaction: low, medium, high)
- Discrete: countable values (e.g., number of purchases)
- Continuous: measurable values with many possible outcomes (e.g. temperature or turnover)
Depending on the type of data, choose an appropriate analysis technique. An AI model that does text classification works differently than a model that predicts sales. And different from a model that recognizes an image or converts speech into text. That's why it's so important to have a good understanding of what your problem is before you start working with AI.
Because let's face it: many organizations want to "do something with AI," but don't know exactly what they want to solve. And then disappointment lurks. Or, even worse: you spend budget on a smart AI solution when you might have just solved the problem with BI. Yet in practice, I often see that organizations already have a lot of valuable data, but that it is scattered across departments or systems. That is precisely where the profit lies: before you start with AI, you can achieve a lot by centralizing and cleaning up your data. Consider a first step like making a data inventory: what data do you actually already have? Where is it located? Who manages what? That may sound boring, but without good data foundations, you are building a house on loose sand. Moreover: by setting this up properly, you immediately work on a data culture within your organization. And that is perhaps even more valuable than any model.
So: do you need BI to get started with AI?
The honest answer: no. You don't necessarily need BI to apply AI. You can develop an AI solution without dashboards or data models. But you do need data as well as a clear picture of the problem you want to solve. In that sense, BI often provides a good foundation: it teaches you to look at your processes, to structure data, to extract insights from what you already know. It makes you look at your numbers in an innovative way. And only then can you think further: "What could I predict?" or "What could I automate?"
Sometimes an issue you thought you'd have to solve with AI can be handled just fine with BI. For example: you want to know which department receives the most complaints. You can see that in a dashboard. No AI required. But do you want to automatically classify complaints and predict which ones lead to customer loss? Then you come more toward AI.
So it's not about the tools, it's about the question. What do you want to solve? What do you want to achieve? And what technology fits that best? So don't let the AI hype fool you. Start with the basics: know your data, understand your processes, ask a sharp question. Whether you then build a dashboard or train an AI model - that is the right order.
