The Four Types of Analytics: Answer Descriptive, Diagnostic, Predictive and Prescriptive with Confidence
At 7:40 AM, a city manager at a delivery app sees orders rising faster than riders can reach the hotspots. A dashboard can tell her orders are up; analytics earns its salary only when it explains why, predicts the next two hours and recommends what to change before customers start cancelling.
- Descriptive analytics answers: “What happened?” It summarizes past and current data through dashboards, reports and KPIs.
- Diagnostic analytics answers: “Why did it happen?” It uses drill-downs, segmentation, variance analysis and root-cause analysis.
- Predictive analytics answers: “What is likely to happen?” It uses statistical models, machine learning and forecasting.
- Prescriptive analytics answers: “What should we do?” It recommends actions using optimization, simulation, business rules or decision models.
- The four types form a ladder: hindsight - insight - foresight - action. Do not jump to AI or ML before clarifying the business decision.
- A strong interview answer uses one business example end to end: sales dropped, discount fatigue caused it, demand will soften next week, so shift budget to high-response segments.
The simplest way to remember analytics is not as four separate tools, but as four levels of decision maturity. Each level asks a sharper business question and demands more judgement.
Core Explanation: The Four Types of Analytics
Analytics converts data into decisions. The difference between the four types is not the software used; it is the business question being answered.
1. Descriptive Analytics - What Happened?
Descriptive analytics summarizes historical or real-time data so managers can see the state of the business. It answers questions like: revenue this month, conversion rate by channel, repeat purchase rate, average delivery time and stockouts by warehouse.
Its strength is clarity. Its limitation is that it does not explain causality. A dashboard showing “sales fell by 12%” is useful, but it is not yet an explanation.
2. Diagnostic Analytics - Why Did It Happen?
Diagnostic analytics looks below the headline number. It separates the total change into drivers - region, segment, product, channel, cohort, price, seasonality or operational failure.
3. Predictive Analytics - What Is Likely to Happen?
Predictive analytics estimates future outcomes using patterns in past and current data. Examples include a demand forecast for next week, a credit default probability, a churn score, a fraud risk model or a lead conversion score.
The key idea: prediction is probabilistic, not magical. A model does not say “this customer will churn with certainty”; it says “this customer has a higher estimated likelihood of churn than others.”
4. Prescriptive Analytics - What Should We Do?
Prescriptive analytics recommends the best action under constraints. It combines predictions with business rules such as budget, stock availability, manpower, service-level targets, fairness, regulatory requirements and risk appetite.
For example, if a model predicts high demand in Bengaluru and lower demand in Pune, prescriptive analytics decides how many riders, SKUs, offers or delivery slots to allocate to each city.
Definitions You Can Say in One Breath
INFORMS: “Analytics is the scientific process of transforming data into insight for making better decisions.”
Worked Example: Same Business Problem, Four Analytics Answers
Suppose a quick-service restaurant chain is planning staffing for Friday evening.
Notice the progression: the same data becomes more useful as it moves from reporting the past to recommending a feasible action.
How to Judge Whether Analytics Is Good
Good analytics is not “a beautiful dashboard” or “a complex model.” It is accurate enough, timely enough and actionable enough to improve a decision. Use these metrics in interviews when asked how you would evaluate an analytics solution.
Case Study: MakeMyTrip Uses Analytics Across the Travel Journey
MakeMyTrip shows how an Indian digital business can use all four analytics types across search, booking, pricing, fulfilment and customer experience.
Situation: Indian travel demand is highly dynamic. Long weekends, school holidays, festivals, weather disruptions, airline fare changes and hotel availability can all shift demand quickly. A travel platform cannot manage this only with static monthly reports.
The move: MakeMyTrip-style travel analytics links customer search behavior, inventory availability, fare movement, booking conversion, cancellation patterns and post-booking support signals. The primary driver is a rich digital transaction trail across the travel journey. Supporting drivers include large-scale search data, supplier integrations, personalization, experimentation and operational feedback from customer service.
Outcome and lesson: The win is not caused by “using AI” alone. The primary driver is converting high-frequency behavioral data into faster commercial decisions, supported by supplier depth, product experimentation and operational execution. The strategic so what: analytics becomes powerful when it covers the full loop - observe, explain, forecast and act.
How AI Changes the Four Types of Analytics
AI does not replace the four-type framework. It accelerates each layer and raises the bar for judgement.
Use NotebookLM or Claude: upload a company annual report, investor presentation and job description, then ask: “Map this company's key decisions into descriptive, diagnostic, predictive and prescriptive analytics. Give me one interview-ready example for each.” Verify all numbers before using them.
Interview Relevance
“Explain the four types of analytics with a business example. How would you decide which type is needed for a problem?”
If the interviewer gives you a vague problem like “sales are declining,” do not immediately propose machine learning. First ask: “Do we need to know what changed, why it changed, what will happen next or what action to take?”
Common Mistake
The mistake: Treating predictive analytics as the final answer. Candidates often say, “We will build a model,” but fail to explain what decision the model improves. Why it costs marks: business analytics is judged by action, not model sophistication. One-line fix: always end with the prescriptive step - who should do what, under which constraints, and how success will be measured.
What to Revise Next
Now that you can separate the four analytics questions, revise the ecosystem around them: how analytics differs from reporting, BI and data science, and which roles build each layer.