Key takeaways
- Analytics runs from descriptive (what happened) to prescriptive (what to do)
- Reliable descriptive reporting delivers more value than early prediction
- Trust in numbers comes from shared definitions, one source and visible comparisons
- Start every analysis from a decision question, with an owner for each indicator
Four levels of analytics
Most institutions get the greatest value from doing descriptive analytics well before investing in prediction.
| Level | Question | Example |
|---|---|---|
| Descriptive | What happened? | Enrolments by programme this cycle |
| Diagnostic | Why did it happen? | Which sources and stages explain lower conversion |
| Predictive | What is likely to happen? | Forecast enrolment range for next month |
| Prescriptive | What should we do? | Suggested follow-ups most likely to help |
Where the data comes from
Institutional analytics draws on admissions and CRM data, student records, attendance and results, learning platform activity, finance and communication logs. The quality of analysis depends on the quality and consistency of these sources, which is why a shared data model matters.
Building trust in numbers
- Agree written definitions for key terms (applicant, enrolled, active)
- Show the definition next to every indicator
- Use one source of truth rather than competing spreadsheets
- Show comparisons, not isolated numbers
- Let people drill down to the underlying records
- Publish known data-quality issues rather than hiding them
From insight to action
Ask a decision question
Start from a decision, not from available data.
Choose a few indicators
Pick measures that inform that decision.
Assign owners
Each indicator has a person who acts on it.
Review on a rhythm
Weekly in busy seasons, monthly otherwise.
Record decisions
Note what was decided and check the effect later.
Frequently asked questions
Should we start with predictive analytics?
Usually not. Reliable descriptive reporting with agreed definitions delivers more value first and is the foundation for prediction.
How many indicators should leadership track?
Few enough to discuss in a meeting, typically eight to twelve, each with an owner.