The Four Types of Data Analytics

Data analytics comes in four types, each answering a progressively harder question: descriptive analytics asks what happened, diagnostic asks why it happened, predictive asks what is likely to happen next, and prescriptive asks what we should do about it. This ladder — sometimes called the analytics maturity model — is one of the most useful mental frameworks a new analyst can carry, because it lets you classify any request the moment you hear it.

Knowing which type a question calls for saves enormous effort. It tells you whether a simple summary will do or whether you need a forecast, and it sets realistic expectations for how hard the work will be. This framing runs through the entire data analytics learning hub.

Descriptive analytics: what happened

Descriptive analytics summarizes the past. It answers questions like "How many orders did we ship last month?" or "What was our average response time?" This is the foundation and the majority of real analyst work — reports, dashboards, KPIs and trend charts all live here. It does not explain causes or predict the future; it establishes the facts.

Most of what you build with spreadsheets and BI tools is descriptive. It sounds basic, but a trustworthy description of reality is genuinely valuable, and getting it right is harder than beginners expect.

Diagnostic analytics: why it happened

Diagnostic analytics digs into causes. Once descriptive analytics shows sales fell in March, diagnostic work asks why: Did a promotion end? Did a competitor cut prices? Did a region underperform? This means slicing the data by different dimensions, comparing segments, and hunting for the factor that explains the change. It leans heavily on techniques like drill-down, segmentation and correlation analysis.

Predictive analytics: what is likely next

Predictive analytics uses historical patterns to estimate the future — forecasting next quarter's demand, predicting which customers may churn, or estimating a delivery time. This is where statistics and machine learning enter. Predictions are never certain; they come with a margin of error, and a good analyst always communicates that uncertainty rather than presenting a forecast as fact.

Prescriptive analytics: what to do

Prescriptive analytics is the most advanced. It recommends actions by combining predictions with optimization or simulation — for example, deciding how much stock to order at each warehouse to minimize cost while avoiding stockouts. It is powerful but comparatively rare, because it requires reliable predictions and a well-defined objective to optimize.

A worked example across the types

Here is one small dataset viewed through the first two analytics types.

import pandas as pd

# illustrative monthly sales sample data
df = pd.DataFrame({
    "month": ["Jan", "Feb", "Mar", "Apr"],
    "sales": [520, 540, 410, 430],
    "promo": [True, True, False, False],
})

# descriptive: what happened
print("Average sales:", df["sales"].mean())
print("Lowest month:", df.loc[df["sales"].idxmin(), "month"])

# diagnostic: why — compare promo vs no promo
print(df.groupby("promo")["sales"].mean())

Expected output:

Average sales: 475.0
Lowest month: Mar
diagnostic:
promo
False    420.0
True     530.0
Name: sales, dtype: float64

Descriptive analytics tells us March was the weakest month. Diagnostic analytics offers a candidate reason: months with a promotion averaged 530 in sales, months without averaged 420. That is a clue about why sales dropped when the promotion ended — exactly the layer diagnostic analytics adds on top of description.

How analysts use the types

In a working role, most requests are descriptive or diagnostic. Leadership wants a reliable dashboard (descriptive) and an explanation when a number moves unexpectedly (diagnostic). Predictive and prescriptive work tends to be project-based and often involves specialized teams. Recognizing which type a stakeholder actually needs — they will rarely name it — is a core analyst skill. Someone asking "will we hit target?" wants prediction; someone asking "why did we miss?" wants diagnosis.

The types also map neatly onto KPIs and metrics: descriptive analytics reports the metric, diagnostic explains its movement, predictive forecasts it, and prescriptive recommends how to influence it.

A practical way to remember the ladder is by the verb each type uses. Descriptive reports, diagnostic explains, predictive forecasts, and prescriptive recommends. When a stakeholder speaks, listen for which verb they really mean: "show me last month" is report, "why did it change" is explain, "what will happen" is forecast, and "what should we do" is recommend. Matching your effort to the verb keeps you from over-building — you do not need a machine-learning model to answer a question that only wanted a well-made summary. Correctly reading the intent behind a request is often more valuable than the analysis itself.

Common mistakes

• Promising prediction when only description is possible. If you lack enough history, an honest summary beats a shaky forecast.
• Skipping diagnosis. Jumping from "sales fell" straight to a fix, without finding the cause, often treats the wrong problem.
• Presenting forecasts as certainties. Predictive results carry error ranges; hiding them misleads decision-makers.
• Overreaching for prescriptive work. Recommending automated actions before your predictions are reliable can do real damage.

In interviews

A frequent interview question is "What are the different types of analytics?" or a scenario like "The CEO asks why revenue dropped — what kind of analysis is that?" Naming the four types and correctly classifying the scenario (that one is diagnostic) shows structured thinking. Interviewers also probe whether you understand that most day-to-day work is descriptive and diagnostic, which signals realistic expectations about the job.

Where this fits in your learning path

The four types give structure to everything else in the data analytics hub. Read what data analytics is first for context, then follow the data analysis process to see how these types play out step by step. Understanding the types also prepares you to define good KPIs and metrics, which are the raw material of descriptive analytics.