Timetabling constrained by real availability, and cohort tracking that shows movement rather than a single snapshot.
Send the file, not a brief. No call required.
Education reporting tends to describe a moment — this term's grades, this year's enrolment — when the useful questions are about movement. Which cohort is diverging, which module drives withdrawal, which timetable constraint is actually binding. That is a different data shape: one row per student per period, rather than one row per student.
The four engagements that come up most often
The columns that carry the argument
| A | B | C | D | E | F | G |
|---|---|---|---|---|---|---|
| 1 | Cohort | Term | Enrolled | Retained | Retention | Avg grade |
| 2 | 2025 intake | T1 | 184 | 184 | 100.0% | 62.4 |
| 3 | 2025 intake | T2 | 184 | 171 | 92.9% | 64.1 |
| 4 | 2025 intake | T3 | 184 | 163 | 88.6% | 65.8 |
| 5 | 2026 intake | T1 | 212 | 212 | 100.0% | 61.0 |
Why each one is written the way it is
=COUNTIFS(Enrol[Cohort],[@Cohort],Enrol[Term],[@Term],Enrol[Status],"Active")/COUNTIFS(Enrol[Cohort],[@Cohort],Enrol[Term],"T1")
Retention against the cohort's own starting size, not against last term. Chained term-on-term rates hide a slow drift that a cohort baseline makes obvious.
=AVERAGEIFS(Grades[Mark],Grades[Module],[@Module],Grades[Term],[@Term])
Module averages by term, so a change in a module is separable from a change in the intake.
=SUMPRODUCT((Timetable[Staff]=[@Staff])*(Timetable[Slot]=[@Slot]))>1
Clash detection. A timetable that is only checked by eye will have exactly one clash, and it will be found in week one.
=[@[Contact hours]]/[@[Staff FTE]]
Teaching load per FTE, the constraint that timetables are usually built without and then rebuilt because of.
A worked case from this sector
A department reported healthy overall retention and could not explain a 12-point drop between first and second year. Restructured by cohort and term, the loss concentrated almost entirely after one second-term module with an average 9 marks below the others. The module was not harder in content; its assessment fell in the same fortnight as two others. Moving one deadline recovered most of the gap in the following intake.
Free, and adaptable to your own data
Specific to this work, not generic
We work with pseudonymised extracts. Cohort and progression analysis needs an anonymous student key, a cohort and a term — not names. That keeps the file outside the scope of most of your data-protection obligations.
No, and it should not. The SRS is the system of record; the spreadsheet is the analysis layer over an extract from it. Anything that has to be authoritative belongs in the SRS.
Both, within limits. Clash detection, load balancing and constraint checking work well in a spreadsheet. Full automated timetable generation across a large institution is a solver problem, and we will say so.