Data Analytics 101: The 4 Types Every Founder and Business Leader Should Know
Most founders, when asked what data analytics means, describe a dashboard. A screen with charts and numbers that shows how the business is performing.
Though that is part of it, it’s the rudimentary part, and treating analytics as synonymous with dashboards is why most companies stay stuck at the level of “looking at what happened” and never reach the level where data starts influencing and supporting proactive decisions.
Data analytics is the process of examining data to draw conclusions, make decisions, and guide action.
At its most foundational, that means looking at what happened.
At its most sophisticated, it means predicting what will happen next and recommending what to do about it.
In between these are two crucial stages that growing companies either rush through or skip entirely.
Here are the four types of analytics, why each one matters, and how they connect to the AI capabilities that build on top of them.
Type 1 - Descriptive Analytics: What Happened?
Descriptive analytics answers the question: What happened in our business over a given period?
This is the most common form of analytics in growing companies.
Revenue last month.
New customers acquired last quarter.
Website traffic trends over the last six months.
Support ticket volume by week.
Employee headcount by department.
Descriptive analytics is retrospective as it tells you about the past. However, it doesn’t provide information on why something happened or whether it will happen again. When done well, it establishes the factual record that everything else builds on.
Most growing companies have some descriptive analytics capability: the monthly revenue report, the weekly sales pipeline review, the quarterly business summary. The question is usually not whether they exist, but whether they are reliable, consistent, and accessible to the people who need them.
The common failure modes are:
reports that take hours to produce manually
metrics that mean different things to different people
Dashboards that nobody fully trusts
Data that lives in disconnected systems, requiring manual reconciliation each time a report is needed.
The AI connection: AI analytics tools can automate descriptive reporting: generating summaries, spotting anomalies, and flagging changes without manual intervention. But they require the same thing that good manual reporting requires: clean, structured, consistently defined data.
Poor descriptive analytics capability is one of the clearest signs that an organization isn’t ready for AI.
Type 2 - Diagnostic Analytics: Why Did It Happen?
Diagnostic analytics goes one level deeper, asking: Why did this happen?
Let’s look at a case where revenue dropped 12% last month. Diagnostic analytics would ask:
Which customer segments drove the decline?
Was it a specific product line?
A particular region?
A cohort of customers acquired through a specific channel?
Was it a one-time event or an underlying trend?
This is the analytical work that most growing companies do reactively, when something goes wrong, and someone needs to figure out (i.e. diagnose) the cause.
The challenge is that reactive diagnostic analysis is slow, expensive, and inconsistent. It typically requires an analyst or the founder to spend hours manually pulling data from multiple systems to identify the pattern.
Mature diagnostic analytics capability means having the data infrastructure in place to answer “why?” questions quickly and reliably, without a multi-day investigation each time an anomaly requires an explanation.
The AI connection: AI tools can significantly assist with diagnostic analytics by identifying patterns across large datasets that would take humans much longer to find, surfacing correlated factors, and generating hypotheses about root causes.
But this only works when the underlying data is clean and integrated. An AI tool trying to diagnose a revenue decline from fragmented, inconsistent data will produce fragmented, inconsistent hypotheses.
Type 3 - Predictive Analytics: What Will Happen?
Predictive analytics uses historical patterns to forecast future outcomes.
What revenue are we likely to generate next quarter based on the current pipeline?
Which customers are showing early signs of churn risk?
Which leads are most likely to convert based on their characteristics and engagement?
What inventory levels do we need to avoid stockouts given seasonal patterns?
Predictive analytics almost always involves a machine learning model: a system trained on historical data to identify patterns that can be generalized to predict future events.
This is also the type of analytics most growing companies are missing. They have descriptive analytics (what happened) and some diagnostic capability (why it happened). But systematic data-driven prediction, such as forecasting that updates automatically, lead scores that recalibrate, churn predictions that flag at-risk accounts weeks before they cancel, is often absent.
The gap matters because prediction is where analytics starts to change decisions rather than just describe them.
Knowing your last quarter’s churn rate tells you what happened.
Knowing which accounts are most likely to churn in the next 60 days tells you where to focus retention efforts right now to preserve revenue.
The AI connection: Predictive analytics is often the first meaningful form of AI adoption for growing companies. The capabilities are mature, the use cases are clear, and the tools are increasingly accessible. The prerequisite, as always, is the data foundation.
A predictive model is only as accurate as the historical data it is trained on.
Type 4 - Prescriptive Analytics: What Should We Do?
Prescriptive analytics is the most advanced type, moving beyond predicting what will happen to recommending what to do about it.
This is where you start shifting the conversation from:
“These five accounts are showing churn signals,” to:
“Here are the three specific actions that have historically been most effective at retaining accounts with this profile, ranked by expected ROI.”
The goal is to move from:
“revenue will likely fall short of target” to:
“Here are the pipeline actions that would close the gap, prioritized by probability of success.”
This is where AI is most valuable for business decision-making: surfacing not just information but recommendations calibrated to your specific context and historical outcomes.
It’s also the most data-intensive form of analytics. Prescriptive systems need to have learned not just what outcomes occurred, but which actions were taken and what resulted. It requires well-maintained, consistently captured action and outcome data, a higher bar than most growing companies currently meet.
The AI connection:
Prescriptive analytics is where AI investment yields the highest ROI, but it builds on a stack (descriptive accuracy, diagnostic capability, predictive foundation) that must be in place first.
Companies that try to build prescriptive analytics without that stack in place typically find the recommendations too generic or too unreliable to act on.
Where most growing companies are in the analytics journey
Most founders and business leaders at growing companies operate primarily in descriptive analytics, with some informal diagnostic capability, and very little systematic predictive or prescriptive capability.
If you are at this stage, don’t take this as criticism but as a starting point. The path from where you are to where AI can genuinely augment your business efforts begins with building each analytical capability in sequence, rather than jumping from basic reporting to advanced AI.
The analytics article pillar on this site is structured around this progression. The early articles build descriptive capability: cleaning data, building dashboards, establishing reliable reporting.
The middle articles develop diagnostic capability: understanding why things happen, automating reporting, building a single source of truth.
The later articles develop predictive capability: forecasting, scoring, churn prediction.
Consequent articles bridge to full AI readiness.
Understanding which type of analytics you currently have and which you’re building toward is one of the clearest ways to know where you stand on your AI readiness journey.
Data analytics hierarchy in practice
Think of the four types as a hierarchy where each level builds on the one below.
You can’t perform good diagnostic work without reliable descriptive data.
You can’t build accurate predictive models without the historical data that diagnostic work helps you maintain and trust.
You can’t get to genuinely useful prescriptive recommendations without the validated predictive models that generate the outcome probabilities that prescriptive systems work from.
Most AI hype skips this hierarchy entirely, and Vendors demo prescriptive AI (“here’s what you should do next”) without acknowledging the three layers of analytical maturity that need to be in place for that recommendation to be trustworthy.
The companies that understand this hierarchy are the ones that evaluate AI tools honestly, invest at the right stage, and get compounding returns rather than repeated disappointment.
Your next step
Download the Data Strategy Checklist to identify which level of the analytics hierarchy your business currently operates at and what foundational work is required to move to the next level.
