LearnGrokUse these prompts in order to move from a raw assessment export to a plan staff can deliver and review. They suit subject teams reviewing a common assessment, and pastoral staff coordinating support across a cohort. Keep the original data source beside the output. The output is a working document, not evidence in its own right.
Key point
Start with the outcomes
A low total score does not tell you what to teach. Select and group pupils using the specific outcome they have not yet secured.
Prepare the source material
Before pasting anything, make one clean copy of the assessment data. Use pupil IDs or initials where possible. Include the assessment date, the scale used and the outcome or question each score represents. A total mark without an outcome-level breakdown can flag a concern, but it cannot reliably tell you which intervention to run.
Gather these documents:
- the assessment export or markbook table;
- the expected outcomes taught before the assessment;
- the assessment questions or mark scheme, if available;
- notes on missing scripts, late arrivals or incomplete assessments;
- the timetable, staffing and room constraints;
- details of support already running.
Do not paste an unfiltered whole-school export if you only need one year group and one subject. Remove columns that do not affect the decision. This reduces accidental errors and makes the audit easier to check.
If a pasted table loses its columns or a file is not read as expected, check the current supported input methods in the xAI documentation. These details can be version-dependent. You can also paste a smaller table with a header row and one pupil per line.
1. Audit before selecting
Run Audit attainment data against outcomes first. Its job is to expose missing values, mismatched scales and outcomes that cannot be traced to a score. Do not skip it because the spreadsheet looks familiar. A teacher may know that a blank means absence, but the output will not know that unless you state it.
Read the Data quality checks table before looking at the priority ranking. Correct the source table where you can. Where you cannot correct it, retain the uncertainty in later decisions. For example, a pupil with an incomplete paper should usually be monitored or reassessed, not automatically placed in the lowest group.
Watch out
Do not treat all blanks as low attainment
A missing mark can mean absence, an incomplete script or a data-entry problem. Those require different actions.
2. Select a manageable shortlist
Use Select pupils using stated criteria after the audit. Write the rule before you run the prompt. A criterion such as “below expected in two named outcomes” is checkable. “Needs extra help” is not.
Set the available places honestly. If there are eight places, ask for eight ranked recommendations rather than requesting a broad list and trying to cut it down later. The Monitor, do not place yet section matters. It keeps pupils with unclear evidence visible without making a false certainty about their need.
Read the evidence for every recommended pupil. The explanation should point to an outcome, score or assessment record you can find in the source. It should not rely on a general label such as “weak pupil” or on an assumed reason for performance.
3. Group by the gap to teach
Use Build groups and intervention sessions when you know who is eligible and what they need to learn. Give it the curriculum content already taught. Otherwise, a session sequence can repeat whole-class teaching or move beyond what the pupils were assessed on.
A useful group has a shared, teachable gap and a workable size. It does not need identical scores. Three pupils who cannot apply the same method may form a stronger group than three pupils with similar totals but different errors.
Check
Check each group against the data
Every pupil in a group should have evidence for the named shared gap. If one does not, move them or record why an individual adjustment is needed.
Check the delivery summary against the actual timetable. Count sessions, staff and group sizes. If the plan requires staff or rooms you do not have, change the scope before it reaches colleagues.
4. Decide what counts as progress
Run Set review points and exit criteria before the first session. Choose a baseline and a later check that both test the same intended outcome. A register shows participation. It does not show whether a pupil can now perform the skill independently.
Set at least one early review point. This is where staff can identify a group that is too broad, a task that is inaccessible or a pupil whose need differs from the original evidence. Record the decision after each review: continue, adapt, intensify, exit or gather more evidence.
5. Give staff one usable summary
Use Prepare a staff intervention summary last. This is the version staff should use during delivery. It should state the group, the teaching focus, the next review and the record to update. Keep the full audit and selection tables with the intervention lead so decisions remain traceable.
Note
Keep interpretation separate from records
Store the original assessment result, the intervention decision and later review evidence as separate fields. This makes it possible to see what changed and why.
How to tell the output is wrong
Check every pupil ID, score and group membership against the source table. Then test the logic in reverse: start with one recommended pupil, find their stated gap, find the assessment evidence, and find the session that teaches that gap. If any link is missing, the plan is not ready.
Also look for suspiciously neat conclusions. A result is unreliable if it silently converts blanks into zeroes, reports a total where only domain scores were supplied, assigns pupils despite missing evidence, or gives dates and staff responsibilities you did not provide. Ask the relevant prompt again with the missing information or correct the source record.
When the plan does not work
Do not keep the same group running simply because it was scheduled. At the first review, return to the pupil record and identify whether the problem is weak baseline evidence, an unsuitable grouping, missed teaching, inaccessible materials or a mismatch between the taught content and the check. Update the plan, then rerun the grouping or review prompt with the new evidence. If the evidence remains unclear, record Needs decision and arrange the appropriate assessment or staff discussion before making a further placement decision.
