7 ways to get more from surveys for schools

Most schools and trusts already run some kind of survey. The common frustrations are familiar to anyone who has sat through a results meeting: the data arrives after the moment to act has passed, two academies used different questions so nobody can compare them, or the report is circulated and nobody is named as responsible for what it says. A survey that cannot change a decision is a cost to pupils' and staff's time, and they notice.

The seven strategies below follow the order in which a leader meets the problem, from deciding what to ask, through collecting reliable data from pupils, staff and families, to comparing academies and following results through to action. Throughout, "belonging" means the extent to which a pupil feels accepted, known and valued in school, and a "cohort" is any group of pupils defined from MIS data, such as a year group or pupils with SEND.

1. Start from the decision, not the questionnaire

Surveys grow long when they are built around topics rather than decisions. A topic invites every interested party to add a question. A decision sets a boundary: if no possible answer would change what the school does, the question does not belong. That is the mechanism that keeps a survey short, which in turn protects completion rates and the quality of what pupils tell you.

For illustration, imagine a secondary school evaluating a new tutor-time model. The decision is whether to keep it, adjust it or replace it after a year. The leadership team asks pupils how far they agree that a tutor knows them well, and that there is an adult in school they could go to. Generic items such as "I am happy at school" are dropped, because a middling score would not tell the school whether to change the tutor model.

Many leaders find a short working sequence helps:

1. Write the decision in one sentence, including who will take it and by when.
2. List every candidate question.
3. Remove any question where you can name no result that would alter your plan.
4. Agree who sees the results, and on what date, before the survey opens.

The usual pitfall is a survey that every department adds to. Each addition feels reasonable, yet the finished questionnaire is long and has no centre. One way to hold the line is to ask each requester which decision their item would inform, and to offer a separate, targeted pulse for anything that does not fit.

Two measures show whether this is working. The first is the share of questions linked to a named decision, which should approach all of them. The second is whether the decision was in fact made using the results, which is the more telling check on whether the survey earned its place in the calendar.

2. Use structured, consistent pupil questions

Pupil data is only as trustworthy as the questions behind it. Short scales with fixed wording and a consistent response format reduce the noise that comes from different readings of the same item, and they let a school say that a score has moved rather than that the question has changed. Pupil surveys also work best with structured items only: agree/disagree statements, frequency scales or multiple choice. They produce clean patterns that can be reported by group without anyone having to interpret individual wording.

Consider a primary school that uses the same frequency scale every term, from "never" to "always", for statements such as "I have someone to play with at break". Younger pupils and those learning English as an additional language complete it independently because every question can be read aloud and translated. The school does not need an adult to explain items, which would otherwise colour the answers.

In practice, schools that do this well tend to settle a fixed question set early and treat it as a standing instrument. They provide read-aloud and translation support as standard rather than for a few pupils, and they tell pupils plainly whether responses are anonymous or named. Each choice has trade-offs. Anonymity can encourage candour, while named responses allow follow-up with particular pupils and cohort analysis from MIS data. Neither is always better, so the important thing is that pupils are told which applies and why.

The pitfalls are rewording and free-text. Tweaking a question each term to make it "better" quietly breaks comparison, so improvements are best saved for a planned annual review. Comment boxes for pupils add safeguarding burden and make results harder to analyse, and a rating scale does the job.

To judge progress, it helps to look at completion rate by cohort and year group, since a strong overall figure can hide a group that is not responding. Stability matters too: for unchanged questions, large unexplained swings between terms often point to a problem with how the survey was run rather than a real change in pupils' experience.

3. Run surveys at a fixed point each term

A single survey is a snapshot. Several taken at the same point in successive terms form a trend, and a trend is what lets a leader tell a seasonal dip from a real problem. A baseline, meaning the first measurement for a group against which later results are compared, matters most for new joiners, because Year 7 belonging in the autumn is a reference point for how transition went.

Imagine a secondary school that runs its baseline in week 6 of the autumn term and repeats it in the same week each term. It avoids mock exam periods because pupils sitting papers respond differently from those in a normal week, and the school wants any change in scores to reflect experience, not timetable pressure. By the summer, it has three comparable points for each year group.

Schools that manage this tend to put the survey windows into the calendar a year ahead, alongside assessment weeks and trips, so they are not squeezed out by competing priorities. They also timetable a 5 to 10 minute slot, often in tutor time, rather than leaving completion to chance, and they run a baseline for pupils who join mid-year.

The most common error is the last week of term. Attendance dips, routines loosen and pupils are tired, so both response rates and the answers themselves are affected. Exam weeks carry the same risk. A fixed mid-term slot avoids both.

Two indicators show whether timing is working. Response rate per window should be steady, and the school should be able to compare like for like across at least three terms. Until there are three points, leaders are generally better off describing movement cautiously.

4. Add staff and parent voice to pupil data

Pupil results on their own tell you how pupils feel. Placing them next to what staff and families say shows where perceptions diverge, and the gaps are often more useful than the scores. A school in which adults believe relationships are strong and pupils disagree has found something no single-audience survey could show.

For illustration, suppose staff rate relationships as strong, yet Year 8 pupils score lower on feeling known by an adult. Rather than debating whose view is right, the leadership team takes the gap to a review of mentoring in Year 8, where contact time had thinned after the transition year.

A shorter parallel set of questions for staff and families keeps burden low and makes matched topics possible. Open-text questions suit adults, and a brief comment box for staff and parents often explains a score in a way the scale cannot. Pupil surveys stay structured. When results are reported, putting groups side by side on matching topics is more informative than separate reports.

The pitfall is running staff and parent surveys as separate projects, owned by different people and reported at different times, so no one connects them to pupil findings. Naming a single lead who reads all three audiences together tends to fix this.

Useful measures include response rates for staff and parents, which are often lower than for pupils and deserve their own attention, and the size of the gap between groups on matched topics. Leaders interested in retention can also track staff turnover over time alongside staff survey results, using DfE school workforce data as the national reference point. Any figure quoted from that source should be checked against the latest release before it goes in a board paper.

5. Break results down by cohort

A school average blends the experience of every pupil, so a group that is struggling can sit unseen behind a healthy headline. Cohort analysis separates those groups out. It matters for accountability as well as practice: the white paper Every Child Achieving and Thriving (February 2026) puts weight on inclusion, which makes the experience of groups such as pupils with SEND a legitimate focus for leaders and trustees.

Imagine a school whose overall score for "adults here listen to me" looks comfortable. Split by cohort, Year 9 pupils with SEND score well below their peers. The average never showed it, because that group is a small share of the year. With the gap visible, the SENCO and head of year can plan a response aimed at the group concerned.

Cohorts are most reliable when generated from MIS data rather than typed by hand or self-reported by pupils. Typical groupings are year group, SEND, Pupil Premium, EAL, FSM and children in care, and it is worth looking at combinations, because a pupil who is both in receipt of Pupil Premium and has SEND may have a different experience from either group alone.

Two pitfalls stand out. One is reporting only averages. The other is the opposite: reporting groups so small that individual pupils could be identified, which undermines trust in the survey. Most schools set a minimum group size below which results are suppressed or merged, and state it openly.

To track progress, it helps to measure the gap between each cohort and the school average, and then follow the lowest-scoring cohort from term to term. If that group is moving up, the response is working. If it is static, the action needs revisiting.

6. Compare academies like for like

Trustees often rely on headteacher summaries, which reflect each leader's judgement and framing. Comparable data from pupils, staff and families gives the board a common reference. Comparison only works when questions, method and timing are identical across every academy, otherwise differences may reflect how the survey was run rather than how pupils feel.

For illustration, picture a six-academy trust. Two academies show an autumn dip in Year 7 belonging. Looking across the trust, the board sees that a third academy has held its Year 7 scores steady, and the trust education director arranges for that academy's transition lead to share practice with the other two.

Setting this up is mostly an exercise in agreement. The trust fixes a single question set and survey window for all academies, and presents trends over time instead of a single snapshot. Comparison with academies in a similar context is fairer than comparison with the trust as a whole, since intake differences shape results. The board also needs to agree in advance how it will use the data, for example to direct support and shared learning.

The main risk is turning comparison into a league table. If heads expect to be ranked, they have every reason to manage the survey rather than use it, and honest engagement drops. Presenting the data as a planning tool, and asking what each academy needs, keeps it useful.

Two measures help. The spread between academies over time shows whether differences are narrowing, and the number of academies with a documented response plan shows whether the data is prompting action.

7. Close the loop with owners and follow-up

Data changes little until someone is responsible for it. The cycle that tends to work is measure, segment, connect, respond and assure, repeated each term. Segmenting finds the low-scoring cohorts, connecting links them to attendance and behaviour data from the MIS, responding assigns action, and assuring gives governors and trustees evidence they can read. Surveys sit alongside existing data rather than replacing it, and the cross-check is where they earn their place.

Imagine a SENCO who takes ownership of an intervention for a low-scoring cohort. The action is specific, and the next termly survey is the review point. The result, whether scores moved or not, is reported to the local governing body in a short note that sets out what was done and what happened.

It helps to follow a simple sequence:

1. Assign one owner and one action to each priority cohort.
2. Set a review date tied to the next survey window.
3. Share a "you said, we did" summary with pupils, staff and parents.
4. Cross-check results against attendance and behaviour data.

The pitfall is sharing a report without owners or actions. Everyone has read it and nobody has been asked to do anything, so the next survey shows no change, and respondents conclude that taking part makes no difference. Response rates then fall.

The measures are the proportion of priority cohorts with an owner and an action, the change in those cohorts' scores next term, and the trend in response rates. Rising participation is often the first sign that pupils and staff believe the survey leads somewhere.

Where to begin, and how Satchel Pulse can help

For most schools the order matters. Fixing the purpose and the timing this term costs nothing and makes everything else more reliable. Adding cohort breakdowns to the baseline comes next, since that is where hidden gaps appear. Naming an owner for every result should come before any move to trust-wide comparison, because comparing academies that cannot yet act on their own data adds reporting without adding improvement.

Satchel Pulse built Perceptions to support this cycle. It provides termly belonging surveys for pupils, staff and families using a consistent question set, with pupil questions kept structured, and distributes them through your MIS. Results are reported automatically by cohort, such as year group, Pupil Premium and SEND, and compared over time and across academies. For staff and parent free-text, AI analysis groups themes and suggests actions. Our own analysis suggests belonging data can flag academies and cohorts at risk of rising absence, suspensions and exclusions up to 18 months before they appear in attendance or behaviour data.

Learn more about our services