To measure SEND intervention impact, compare a suitable baseline with the same or a genuinely comparable outcome measure, then check what was actually delivered, whether the pupil could access it, whether learning transferred beyond the session and what the pupil and family observed.
A rising graph is evidence to examine, not a verdict. A flat graph may reflect limited impact, incomplete delivery, an unsuitable measure or a change that the measure cannot capture.
This guide is for schools in England and is accurate on 8 August 2026. It supports professional review, not diagnosis, legal advice or a claim that one intervention caused an outcome.
Why review intervention impact now?
Summer planning gives a SENCo time to decide which support should continue, change or stop before the new timetable hardens around it. A graph can make that decision look pleasantly scientific, even when half its sessions disappeared beneath trips, absence and room changes.
England's SEND Code of Practice says schools should review the effectiveness of support and its impact on progress at the agreed date, with the pupil and parents involved. The review sits within assess, plan, do and review, so it should improve the next cycle rather than merely close the previous one.
The Education Endowment Foundation's implementation guidance separates pupil outcomes from implementation outcomes such as fidelity, reach, acceptability and feasibility. That distinction matters: an intervention can be promising in principle but poorly delivered locally, or delivered faithfully without producing a worthwhile change for this pupil.
Ofsted's inspection information for use from September 2026 focuses on the quality and impact of leaders' decisions, while saying schools do not need to manufacture written evidence for each toolkit standard. No material official development in the previous 72 hours changed the workable position.
Use four evidence layers, not one progress line
Start by naming the decision. Are staff deciding whether to continue, adapt, step down, stop or reassess? Then assemble four layers of evidence that can answer that decision without asking one measure to do every job.
| Evidence layer | Record | Review question |
|---|---|---|
| Baseline and outcome | Measure, date, context, assessor and intended change | Is the comparison valid and educationally meaningful? |
| Dosage and fidelity | Planned and delivered sessions, duration, group size and approach | Was the intended support actually delivered? |
| Access and participation | Attendance, missed-session reasons, engagement and adjustments | Could the pupil participate sufficiently for a fair review? |
| Transfer and views | Classroom use, work samples, observations, pupil and family views | Did the change matter beyond the intervention task? |
Check the baseline before interpreting change
Record the starting measure before delivery where possible. Include the date, task conditions, person recording it and relevant context. A score collected after two introductory sessions is not a clean baseline, although it may still be usable if the limitation is explicit.
Use the same measure at review, or explain why a replacement is comparable. Check for floor and ceiling effects, changes in difficulty, altered scoring rules and practice effects. Three dots joined by a line remain three observations; software does not grant them a larger sample size.
Define what would count as a worthwhile change before seeing the result. Consider educational significance; it is not only numerical movement. A small score change may matter if the pupil now uses the skill independently in class, while a larger task score may have little value if it never transfers.
Audit dosage and attendance before judging effectiveness
Compare planned sessions with delivered sessions. Record cancellations, shortened sessions, staff substitutions, changes in group size and material departures from the intended approach. This is implementation evidence rather than an excuse column.
Separate school cancellation from pupil absence and from attendance at school but not at the session. Each suggests a different professional question. Also check whether the intervention itself created an access problem, such as repeatedly removing the pupil from a valued subject.
Low exposure usually limits the conclusion. It may be reasonable to say there is not enough delivered evidence to judge impact. Do not convert that uncertainty into “no progress”, and do not automatically prescribe more of the same without reviewing fit and feasibility.
Read graphs as prompts for questions
Keep axes, intervals and scales consistent. Show missing observations honestly. Avoid a chart that starts just below the first score merely to make a small movement look mountainous. The mountain will not improve because the y-axis has become ambitious.
Look for patterns that deserve follow-up: rapid early gain followed by a plateau, variation by context, improvement during a period of high attendance, or a change that appears only on the practised task. Compare these with notes and work; it is not with a desired narrative.
For small groups and individual pupils, resist causal language. School records rarely provide a credible counterfactual. Report what changed, what was delivered, other plausible influences and the confidence staff place in the judgement.
Test the pattern against its context
Picture a fictional composite pupil receiving an eight-week reading intervention. The progress graph is almost flat, so the first signal appears to be limited impact.
The session record shows that only nine of sixteen planned sessions were delivered. Attendance data explains three missed sessions, while timetable notes show four were cancelled by the school. Classroom work and teacher observation show the pupil now uses one decoding strategy in lessons, although reading fluency remains below the agreed outcome.
The professional question changes from “Did the intervention work?” to “What can nine delivered sessions tell us, what transferred, and can the school provide an accessible next cycle with a measure better matched to the intended change?”
The action is to fix the delivery barrier, retain the helpful classroom strategy, agree a new review point and decide whether to adapt or replace the programme. The combined records do not prove that sixteen sessions would have produced the intended outcome.
What remains unknown
A before-and-after score alone cannot show that the intervention caused the change. It cannot diagnose a need, decide SEN status, amend an EHC plan or establish that a pupil lacks effort.
Attendance data records access rather than the reason for every absence or the quality of participation. Delivery counts do not prove fidelity or a positive experience. Pupil, family and staff views add essential context, but they are not interchangeable with a validated outcome measure.
Where evidence conflicts, record the disagreement and lower confidence in the conclusion. A fair review can end with uncertainty and a plan to gather better evidence. That is more useful than a confident answer assembled from incompatible measures.
Using Student Radar for the review
In Student Radar, the Intervention Planner keeps planned sessions, delivered sessions, attendance, notes, success criteria and a review date together. Outcome Tracking records evidence, confidence, a professional judgement and the next action beside SEND targets.
Where attainment is the relevant outcome, Assessment Reports can provide school, cohort or pupil context. These views support human review. They do not attribute causation or decide whether support should continue automatically.
A short route into action
- Name the decision each intervention review must support.
- Check baseline and outcome measures are genuinely comparable.
- Reconcile planned dosage with sessions actually delivered.
- Separate pupil absence, school cancellation and access barriers.
- Add classroom transfer, pupil views and family views.
- Record alternative explanations and confidence in the judgement.
- Choose continue, adapt, step down, stop or reassess, with an owner and date.
- Use the review learning to improve the next assess, plan, do and review cycle.
Pair this review with the intervention entry and exit criteria guide. For leadership reporting, use the SEND provision impact report model without turning an individual review into a league table.
To see the connected workflow in context, request a focused Student Radar walkthrough.
Sources and further reading
- SEND code of practice: 0 to 25 years, Department for Education and Department of Health and Social Care, published 11 June 2014; GOV.UK page updated 12 September 2024.
- A School's Guide to Implementation, Education Endowment Foundation, third edition published 24 April 2024.
- Guide to Inclusive Teaching: adaptations and additional support, Education Endowment Foundation, published 6 July 2026.
- Inclusive mainstream fund: best practice for schools, Department for Education, published 25 June 2026.
- Inspection information for state-funded schools: for use from September 2026, Ofsted, published 12 June 2026; applies to inspections from September 2026.