The IDSR is full of data. What it can't do is tell you what to say when someone asks you about it. You can read it twice and still not know which few lines will come up in conversation with an inspector, or what you'd point to in reply.

We've been working on that this week. We've written a prompt that takes a primary IDSR and turns it into a short planning document. It sets out strengths you can evidence and risks an inspector may raise. For each risk you get the question likely to sit behind it and the evidence to have ready.

Before anything else: where the data goes

The IDSR isn't a public document. Where cohorts are small, some figures could point to individual pupils. Please don't paste it into a free AI tool you've signed up to personally. Use one your school or trust has approved for school data. Check with your data protection officer if you're not sure. Remove anything that could identify a child before you start. If your school has no approved tool yet, the prompt is still useful as a checklist to work through by hand.

Why the rules matter more than the request

Ask an AI assistant to summarise a data report and it can produce something that sounds confident and is wrong in the details. Numbers get rounded, results get called significant when the report doesn't say so, and small groups get treated as if they were large ones. Most of our effort went into rules that stop this. These are the ones worth keeping even if you rewrite everything else:

  • Use only figures that appear in the IDSR, and quote the number, the year and the comparison. If something is missing, redacted or suppressed, say so and don't guess.
  • Use the report's own wording for bands and significance. Nothing is significant unless the report says it is.
  • Be careful with small groups, and flag when one pupil moves a percentage a long way.
  • Don't predict a grade. The IDSR is a starting point for a conversation, and the prompt treats it that way.
  • Keep what the data shows separate from what it might mean, and label any explanation as a hypothesis for you to test.

The last rule matters most to us. It keeps the AI from writing your school's story for you. You decide what the numbers mean, because you know the children behind them.

What it gives you back

The prompt works through the report section by section, covering context, attendance and behaviour, phonics and KS2 attainment, progress, disadvantaged gaps, and trust context if you're part of one. If your report doesn't include a section the prompt asks about, the first rule tells the AI to say so and move on.

You then get the three strongest positives you can evidence and the three biggest risks. For each risk there's the data point, a likely inspector question and the evidence you'd want to show. The prompt also looks for patterns across sections, such as high persistent absence in a year group that also shows weaker results. It lists what the IDSR can't tell you, like in-year tracking or the detail of your SEND provision, and ends with ten questions to prepare for and a one-paragraph summary you could say aloud in under a minute, using only points you can back up.

Before you use it

Check every figure the AI quotes against your report before it goes into a meeting or your SEF. Treat the output as a first pass, and add your own internal data and context before you talk to anyone outside school.

The prompt was written for primary schools and follows the layout of the primary IDSR. Secondary colleagues would need to adapt the attainment and progress sections. If your report looks different from what the prompt expects, adjust the prompt to match the report.

Where to find it

The full prompt and a seven-section template are in the IDSR Pack, in the members' area of our community. Each section of the template is a table with a "So what?" column, so every number leads to a point you can make.

Join HeadteacherChat to get the prompt and template

Jonathan