Most school management systems demonstrate well. An hour with a salesperson and a clean demo database will make almost any of them look capable. The differences that matter show up in month four, when the register has to be taken on a Tuesday morning in a block with no signal, when a transitional cohort needs a second grading scale, and when the bursar asks why the arrears list disagrees with the ledger.
These are the questions worth asking before you sign, roughly in the order they tend to hurt.
On curriculum
1. Can the system run two curricula at once, in one school?
Not “does it support CBC” — every vendor says yes. Ask whether CBC and KCSE can operate concurrently in the same school, each with its own ladder of levels, its own subject availability and its own grading scale. Schools in the middle of a transition need exactly this, and a system that assumes one curriculum per school forces a second installation or a set of workarounds nobody documents.
2. Who defines the grade boundaries?
If the answer involves the vendor making a change for you, the system will be wrong for your school within a year. Grade ranges, points, remarks, bands and the pass mark should be things your deputy configures on a Tuesday afternoon.
3. What happens to historic results when boundaries change?
Raw marks and reported grades should be separate. Changing a boundary should re-report results without altering a single learner’s mark. If revising a boundary means re-entering marks, the school will simply never revise a boundary.
On the people who actually enter the data
4. Where does a teacher take the register?
If the honest answer is “on paper, and the office types it up”, you are not buying an attendance system. You are buying a place to store attendance that somebody else will retype — with the delay and the transcription errors that implies. Attendance is only actionable on the morning it is taken.
5. Does it work without a connection?
Ask specifically about registers and mark entry. Connectivity in a classroom block is not the same as connectivity in the administration office, and a system that stalls without a signal is a system teachers stop using inside a term.
6. Who enters marks?
The teacher who did the marking should enter them, into a mark sheet scoped to the subjects and streams they actually teach. Every retyping step between the red pen and the report card is a step where errors enter and time is lost twice.
On parents
7. What reaches a parent without a smartphone?
This is the question that most cleanly separates systems designed for Kenyan schools from systems designed elsewhere and sold here. A meaningful proportion of guardians do not use a smartphone, and they are disproportionately the families where school contact matters most.
Ask whether absence alerts, fee reminders and results go out by SMS automatically — not whether SMS is available as a paid extra somebody has to remember to trigger.
8. Is a message an action, or an automation?
The messages that matter most should not be composed by anybody. An unexplained absence should notify a guardian because a learner was marked absent, not because an administrator sat down to write to the parents of absentees.
9. Does a guardian with three children have three logins?
If yes, the parent app will not be used. Families are not tidy: children are in different classes, sometimes at different schools, and guardianship arrangements vary.
On money
10. Where does the fee balance come from?
A balance should be derived from invoices and receipts, not stored as a number somebody maintains. If it is stored, it will drift, and the arrears list, the reminder sent to the guardian and the figure in the principal’s report will eventually disagree.
11. How are M-Pesa payments matched?
Ask to see the reconciliation. The realistic goal is not that every payment matches automatically; it is that the unmatched ones are a short exception list rather than a search through a statement.
On leaving
12. How do we get our data out?
Ask for a demonstration of the export, not a policy statement. Every list — learners, staff, marks, attendance, fees — should export to a usable format, on demand, without a support ticket. A vendor confident in their product has no reason to make leaving difficult.
None of these questions are exotic. They are simply the ones that get skipped in a demonstration, because a demonstration runs on a fast connection, with one curriculum, a tidy family structure and a database nobody has to leave.