A form in March, signed and filed, that nothing follows from. Here is why school appraisals fail in the same four ways every year, and what has to exist before a rating means anything.
Most schools have tried an appraisal. Usually it is a form, circulated in March, filled in over a weekend, signed, and filed. The next conversation about it happens a year later.
It fails in the same four ways every time, and none of them is fixed by a better form.
Nobody agreed what good looks like. "Teaching quality" means something different to the person being rated and the person doing the rating. Without a written measure and a target set in advance, a rating is an opinion — and an opinion cannot be discussed without it becoming personal.
It is one event a year. A year is long enough for everybody to forget what happened in term one, so the rating reflects the last six weeks. Staff know this perfectly well, which is why visible effort rises in February.
The form is where it ends. A score is produced and nothing follows from it. No training is arranged, no decision is tied to it, nothing changes. By the second year people fill it in without reading it, which is the correct response to a ritual.
The training plan is a wish list. Somebody asks heads of department what their people need. The answers are reasonable guesses, not measurements, and the budget goes to whoever asked most persuasively.
Three mechanisms fix all four, and they only work together: a written measure agreed in advance, a cycle shorter than a year, and a skills record that produces the training plan as a by-product rather than as an opinion.
A KRA is one thing a job is answerable for, with a number attached. You write them per role, not per person, and reuse them every cycle.
Each carries five things: a title in the school's own words, a weight as a percentage of the total, a measure (a number, a percentage, an amount, a rating out of five, or a yes/no), a target, and a frequency.
Here is a complete set for a class teacher:
| KRA | Measure | Target | Weight |
|---|---|---|---|
| Syllabus completed to plan | Yes / No | Yes | 20 |
| Own attendance | Percent | 95 | 15 |
| Parent meetings held against planned | Percent | 100 | 15 |
| Assessment records submitted on time | Percent | 100 | 20 |
| Head of department's rating on classroom practice | Rating / 5 | 4 | 30 |
Five lines. Every one answerable with a fact or a mark rather than a feeling. The weights are how the school says what matters most — they do not have to be equal, and the argument about the weights is a far more useful argument than the one about the final score.
Choosing the measure honestly is most of the work. "Completes the syllabus on time" is a yes/no, and pretending otherwise invents precision. "Parent meeting attendance" genuinely is a percentage.
And where the system already holds the number — attendance, for instance — it fills itself in. The person can type over it, but the original stays visible beside the override, because a figure quietly replaced is worse than no figure.
Open a cycle, and the right KRAs are assigned to every member of staff automatically, based on their designation and department. Then four steps:
That last row is what stops a cycle dying quietly. The school finds out it is stalling while it can still be rescued, rather than in the week it was supposed to conclude.
Run it quarterly and the conversation is about the last twelve weeks, which everybody can actually remember.
Achievement is achieved ÷ target, capped at 100. The cap is deliberate: over-delivery on one KRA cannot pay for a miss on another, because a teacher who held twice the planned parent meetings has not thereby completed the syllabus. If the school wants one thing to count more, that is what the weights are for.
The weighted total is taken across every KRA, including any left unscored, which count as zero — and the sheet says how much of the weight the figure actually rests on. A sheet where four of six were scored does not get to present itself as a complete result.
It is the difference between a number you can take into a confirmation meeting and one you cannot.
This is the mechanism schools most often skip, and it is the one that makes the whole thing lead somewhere.
A skills matrix is a grid: people down the side, competencies across the top, and a colour where somebody is below what their role needs. Course completions tell you what was delivered. This tells you what the school can do.
Two things keep it honest.
The required level comes from the role, not the person. You set what a designation needs — spoken English at a given level for class teachers, spreadsheet skill at the accounts desk, first aid for sports staff. A gap is therefore a fact about the role's requirement, not an accusation about the individual, and that distinction is exactly what lets a head of department raise it without the conversation going wrong.
Skills go stale. A competency can carry a decay period, after which a level with no fresh evidence counts at half weight. Without this, "trained in 2019" reads identically to "trained last month" — the most common lie in training records.
Now filter the matrix to gaps only. That list is the training plan: measured rather than guessed, and already grouped by department and designation, which is to say already grouped into the batches you would run. It took no survey to produce.
Feedback collected immediately after a session tells you whether people enjoyed it. That is the easiest thing to measure and the least useful.
What matters is whether behaviour changed, and that is a separate, later check: a short structured question to the person's manager some weeks afterwards, asking whether the thing trained is now visible in the work. Put the cost of the training next to that answer and, after two cycles, the school stops arguing about training budgets because it has evidence instead of advocacy.
Not with the whole thing.
Write KRA templates for one designation, not the whole school. Run one short cycle — a quarter, not a year — end to end, and close it. You will change the templates after the first cycle; changing five is much cheaper than changing fifty.
Then build the skills matrix, filter to gaps, and run the batches it produces. Each of those phases is useful on its own even if the next is delayed, which matters, because the schools that succeed at this are the ones that shipped something small in the first term rather than designing something complete for the next academic year.
The first cycle that produces no consequence is the one that teaches everybody to stop taking it seriously. So before you open a cycle, decide what will follow from it.
A companion piece covers the structure underneath all of this: why your school's org chart needs to be a field, not an understanding.
A 30-minute working session on your own designations is usually enough to write the first templates and open a trial cycle.