Not all data analysis answers the same question. Some tell what happened, others what will happen, others what we should do. Understanding the three levels of analytics — descriptive, predictive and prescriptive — helps you know where your company stands and where it can evolve.
Descriptive analytics: what happened
It is the starting point and the most common: reports, dashboards and KPIs that summarize the past. "We sold X last month", "the North grew 10%". It answers what happened and is the basis of everything — without a good description of the present, there is no reliable prediction.
Predictive analytics: what will happen
Here we use historical data to anticipate the future: forecast next quarter's sales, estimate which customers are at risk of leaving, project demand. It rests on statistical and machine learning models that learn patterns from the past and project them forward.
Prescriptive analytics: what we should do
The most advanced level not only predicts but recommends actions. "Since these customers will leave, offer this discount"; "to minimize transport costs, use these routes". It combines prediction with optimization and business rules to suggest the best decision.
A ladder, not a choice
- Descriptive: essential, where almost everyone starts.
- Predictive: requires clean historical data and some analytical maturity.
- Prescriptive: demands the two previous ones well built, plus optimization.
The mistake of skipping steps
Many companies want "AI that decides" (prescriptive) without having descriptive in order — without trusting the numbers they already have. It is like wanting to run before walking. Value grows level by level, and each rests on the previous one.
In practice
Find out which level you are at: are your reports reliable? Do you predict anything yet? Do you automate decisions yet? Climbing one step at a time is safer and more profitable than jumping to the trend. Which level of analytics is your organization at today?
