How EEF implementation guidance applies to belonging work

A trust picks a belonging priority in September, agrees a plan at the autumn board, and by February the intervention has quietly stalled. Tutors have not had the time, the lead has moved on to something else, and nobody can say whether the approach was ever delivered as designed. The EEF's implementation guidance explains why good ideas fail at this point, and it gives belonging work a structure that perception data can support at every phase.

The problem is rarely the choice of priority. It is the gap between choosing and embedding, which is where most of the effort, and most of the risk, sits. The sections below take the four phases in turn, with a worked trust example.

What the EEF says about why good ideas fail in schools

The Education Endowment Foundation's guidance, Putting Evidence to Work: A School's Guide to Implementation, was first published in 2018 and has been updated since. Its central message is that how a programme is introduced affects whether it works as much as what the programme is. A well-evidenced approach delivered patchily, without training or follow-up, can produce little. A modest approach that is planned, supported and checked often does better. Readers can find the current version on the EEF website, and it is worth checking the latest wording there.

The guidance treats implementation as a process that runs over months, not an event. That matters because three terms are often used as if they meant the same thing.

• An intervention is the programme, practice or routine itself, such as a mentoring scheme or a transition programme.
• Adoption is the decision to take it up, which usually means a purchase, a policy or a line in the improvement plan.
• Implementation is everything that happens between that decision and the point where the practice is part of how the school works.

Buying a programme is adoption. Embedding it, so that it runs the same way in October and in May regardless of who is on duty, is implementation.

The most common misconception is that a launch counts as implementation. A well-run INSET session or a single baseline survey feels like progress, and both are useful, but each is one moment in a long sequence. The EEF describes four phases, explore, prepare, deliver and sustain, with monitoring running through all of them. Schools that stall in February have often completed the first moments of the process thoroughly and then had no structure for the rest.

The four phases and the questions each one asks

Each phase of the EEF model can be read as a question the school needs to be able to answer before it moves on. The wording below is a paraphrase, and the current publication gives the exact recommendations.

• Explore asks whether this is the right problem, whether the chosen approach fits it, and whether the school is ready to take it on.
• Prepare asks whether there is a clear plan, trained staff, the necessary resources and a named implementation team to run it.
• Deliver asks whether staff are supported, whether progress is being monitored, and whether problems are being solved and the approach adapted as it meets real timetables.
• Sustain asks whether the practice is embedded, and whether the school will review it and decide what happens next.

The implementation team deserves a note, because it is where many belonging priorities lose momentum. The guidance points to a small group with the authority and time to make decisions, led by someone who can link day-to-day delivery to senior leadership. In a trust, that might be an academy lead and a trust inclusion director, not a single enthusiastic teacher.

Active ingredients and the line between adaptation and drift

The EEF also uses the idea of active ingredients, the elements of an approach that are believed to make it work. Separating these from the parts that can flex gives schools permission to adapt without losing the point. If a mentoring routine works because every pupil has a regular, named adult and a consistent check-in, then the day and the room can move. The named adult and the regularity cannot.

This is the distinction between fidelity and adaptation. Fidelity means delivering the active ingredients as designed. Adaptation means changing the surface details to fit the school. Drift happens when adaptation quietly removes an active ingredient, and nobody notices because nobody defined it in the first place. Agreeing the active ingredients during preparation is what makes later adaptation a deliberate choice and not an accident.

Where perception data belongs in the explore phase

Explore begins with diagnosing the need, and this is where the evidence available to leaders is often thinnest. Attendance figures, suspension data and headteacher summaries describe outcomes, and they arrive after the experience that produced them. By the time a pattern is visible in absence data, pupils have often felt a certain way about school for some time. Pupil, staff and parent perceptions show what people are experiencing now, which makes them a better basis for deciding what problem the priority should address.

Segmentation changes what the data says. A trust-wide average of belonging can look healthy while one group sits well below it. Many trusts read results by year group, Pupil Premium or disadvantaged status and SEND, and then by combinations of these. A finding such as "Year 7 pupils with SEND were less likely to agree that they have somewhere to go at lunch" points to a specific problem and a specific group, which is a much better starting point than a general aim to improve belonging. Pupil questions in this kind of work are best kept structured, with rating scales or agree and disagree statements, so that results are comparable and every pupil answers the same question.

Readiness is a perception question too

The second half of explore is readiness, and it often gets less attention than diagnosis. A school can identify the right problem and the right approach and still not be ready to deliver it. A short staff survey before commitment can test this. Typical items cover whether staff feel they have capacity for a new routine, whether current workload is manageable, whether they understand why the change is proposed, and whether they believe it will help pupils.

Low readiness scores are not a reason to drop the priority. They are planning information. A school where staff report little capacity may need to remove something before adding something, or phase the rollout. A school where staff support the aim but doubt the method may need more explanation, or a different approach. Finding this out in October is far cheaper than finding it out in February.

A worked example: Year 7 belonging across three academies

The following is an illustration, not a real result. Suppose a trust of three secondary academies runs the same belonging survey for Year 7 pupils in the autumn term, with identical questions and the same fortnight for completion. Two academies score broadly in line with each other. In the third, Pupil Premium pupils score noticeably lower than their peers on items about having an adult to talk to and feeling that teachers know them. Because the method and timing match, the trust can see this gap is not an artefact of how the survey was run.

Preparing the response

The trust and the academy agree a single owner, in this case the assistant headteacher for pastoral and inclusion, who reports to the trust education director. They choose two active ingredients for a structured mentoring routine: every Year 7 pupil has a named tutor, and that tutor holds a short, scheduled one-to-one conversation with each tutee once a fortnight. Everything else, including the day, the room and the format of the conversation, is left to the academy. They also agree what a good outcome looks like, for example that Pupil Premium pupils' agreement with the "adult to talk to" item moves closer to the cohort average by the spring recheck, and they put the date in the calendar.

Delivering and adapting

Six weeks in, the lead runs a short staff pulse of a few items, asking tutors whether they have been able to hold the conversations as planned, whether they feel confident in them, and whether timetabling gets in the way. Suppose the results show that tutors support the routine, but many cannot fit the conversations into a 20-minute registration slot. The tempting response is to conclude the approach does not work and drop it. The better reading is that an adaptable detail, the timing, is blocking an active ingredient, the regular conversation. The academy moves some conversations to a protected slot in the weekly timetable and keeps the routine itself intact.

In this example, the staff pulse told the team something the outcome data could not have told them for months, which is that the plan was not being delivered as designed.

How to judge whether implementation is working before outcomes move

The EEF's guidance stresses monitoring, and in practice that means two kinds of measure. Implementation measures ask whether the approach is happening as intended. Did tutors have the training, did the conversations take place, did staff have the time? Outcome measures ask whether pupils' experience and behaviour are changing, such as belonging scores, attendance and suspensions. Outcome data alone is hard to interpret, because flat results could mean the approach does not work or that it has not been delivered. Implementation data separates the two.

Repeating the same survey each term makes change comparable. A single snapshot shows where a school stands, but it cannot show direction. If the autumn baseline and the spring recheck use the same questions, the same year groups and similar timing, a movement in a cohort's score means something. If the wording or the window changes, the comparison weakens. A termly rhythm also fits school planning, with results available before each set of leadership and trust meetings.

Some caution is needed in how results are read. Belonging is a leading indicator, which means it tends to move before attendance and behaviour figures do, but it does not guarantee that outcomes will follow. A rise in belonging scores is encouraging evidence that the approach is landing. It is not proof that attendance will improve, and a flat term is not proof of failure.

How results reach trustees shapes what happens next. Presented as a scorecard, a low score invites defensiveness. Presented as a planning tool, with a named owner and a clear next decision, the same score invites a useful conversation. This is what the sustain phase asks for in practice, a decision at each review to continue, adapt or stop. Stopping an approach that is not working, or has done its job, is a legitimate outcome and often a sign of a healthy process.

How Satchel Pulse can help

Implementation quality decides whether a belonging priority works, and perception data helps at the points where leaders otherwise rely on impressions. Perceptions, the survey platform from Satchel Pulse, runs termly belonging surveys for pupils, staff and families, distributed through the school's MIS. Results can be broken down by year group, Pupil Premium and SEND, and compared across academies and over time, which supports the explore phase and the monitoring that follows. Satchel Pulse's own analysis suggests belonging data can flag cohorts at risk of rising absence and exclusions up to 18 months before they appear in attendance and behaviour figures, which is why a baseline is worth having early.

A practical next step this term is to choose one belonging priority and run a baseline survey before half-term planning. Naming an owner and booking the spring recheck at the same time means the plan has a date and a person attached before the autumn ends. Learn more about our services