Reducing disadvantage with background progress analysis

Disadvantage is rarely a surprise by the end of the year. It is a surprise in October, when nobody has the time to notice it, and obvious in June, when it is too late to do much about it. The gap between a pupil eligible for the pupil premium and their peers is usually visible in small things first: a homework not started, a piece of independent work that never matches the quality of the classwork, a hand that stops going up. None of those things arrive as data. They arrive as impressions, spread across five or six teachers who never meet.

Background progress analysis is the attempt to close that lag. It is what the Educational Intelligence Layer (EI layer) - the school-governed layer that sits across planning, teaching, assessment, inclusion and family contact - does quietly, using the evidence a school already produces, so that the pupils who are slipping are named in week three rather than week twelve.

What does "background analysis" actually mean?

It means no new data entry. The EI layer reads what already exists in the school: marks recorded against class assessments, marked work and the comments made on it, homework set and homework completed, practice attempts, and the behaviour and pastoral events staff log as part of their normal week. Those become signals attached to a pupil, and signals accumulate.

The important word is background. Teachers are not asked to fill in a tracker so the analysis can happen. The analysis happens because the teaching happened. If a school adds nothing to a teacher's workload and still surfaces the right five pupils on a Monday morning, it has bought time rather than spent it.

Why continuous analysis beats the termly data drop

A data drop is a photograph. It tells you where a pupil stood on one day, after the term that produced the result has finished. Continuous analysis is closer to a trend line, and trends are actionable while a term is still running.

Two examples of what only shows up continuously:

• Quiet non-completion. A pupil who submits four homeworks out of nine still has attainment data that looks acceptable. The completion pattern is the earlier and better warning.
• The classwork-to-independent gap. Where supported work is strong and independent work is weak, the pupil has not secured the learning - they have borrowed the scaffold. That comparison is invisible in a single grade.

Both are disproportionately common among disadvantaged pupils, and both are cheap to fix early and expensive to fix late.

What it surfaces that a single teacher cannot see

The most useful thing a school-wide layer does for equity is cross-subject pattern recognition. A pupil whose written extended answers are weak in history, geography and science does not have three subject problems; they have one writing problem being managed three times, badly, by three people who each assume it is specific to them.

The tracking view puts those subject signals side by side, so the shared cause is visible and can be addressed once, deliberately, rather than absorbed into three sets of marking comments. It also flags the pupils whose participation is narrowing without any grade having moved yet - the pattern that most often precedes disengagement.

From signal to action: a small number of next steps

Analysis that produces a long list of targets produces nothing. Attention is the scarcest thing in a classroom, so the output is deliberately small: Leverage Action Points, the two or three next steps with the largest likely return for that pupil right now, each shown with the evidence that put it there. No evidence, no action point. Nothing is inferred from a pupil's background, postcode or pupil premium status - the analysis reads work and behaviour, never demographics.

Those action points then travel into teaching. When a teacher selects a class in the planning module, the current action points for that class come with it, and the lesson comes back containing concrete in-lesson moves for those pupils - within the subject being taught. An English lesson stays an English lesson. A pupil's numeracy target does not become a graph task inside a writing lesson.

Adaptability, not lowered expectations

The failure mode of any disadvantage strategy is a quieter curriculum for the pupils who need the loudest one. The rule in the EI layer is one ambitious outcome for the whole class, with the scaffolding adapted and then faded. Sentence stems, worked examples, partial models and structured talk are supports that are designed to be withdrawn, and the plan says when.

This matters most for pupils who arrive with less at home. If the adult response to a gap is to reduce the demand, the gap becomes the ceiling. If the response is to hold the demand and build the route to it, the gap is something a pupil can travel through.

What leaders see

For a pupil premium lead or an SLT member, the value is being able to ask a question and get an evidenced answer in the same sitting: which disadvantaged pupils in Year 8 have declining completion across more than one subject, and what is being done in lessons about it. Year and school views in the leadership tools roll the same signals upwards, and every headline can be opened to the underlying evidence, so a conversation with a head of department starts from work rather than from a colour on a spreadsheet.

Where a pupil also has additional needs, the same signals feed the SEND view, so the inclusion picture and the disadvantage picture are not two separate exercises with two separate sets of paperwork.

Guardrails

Analysis of children's progress has to be governed, not just clever.

• Teacher in the loop. Nothing is sent home, recorded on a plan or acted on formally without a member of staff reviewing it. The layer proposes; the school decides.
• Evidence or silence. If the evidence is thin, the layer says so rather than inventing a judgement.
• Named pupils only where permitted. Pupils are handled as anonymised identifiers unless the school's data agreement is in place and permits real names.
• No training on school data. School content and pupil evidence are used to answer the school's own questions, and are not used to train models.

Does this replace professional judgement?

No, and it should not try to. The analysis is good at the thing teachers cannot do at scale - holding six subjects, twelve weeks and thirty pupils in view at once - and poor at the thing teachers do well, which is knowing why a particular child was different this week. The useful division of labour is that the layer narrows attention and the teacher decides what to do with it.

How long before a school sees anything?

Signals begin accumulating from the first marks and the first homework cycle, so the first credible headlines usually appear within a few weeks of normal use, and sharpen from there. Nothing needs to be backfilled first.

What does a school need to have in place?

Marks or marked work being recorded somewhere in the platform, classes with pupils in them, and a data agreement if real names are to be used. Schools exploring the model can see the whole picture, including leadership views, on the schools page, and the wider argument for a joined-up layer rather than a shelf of separate tools is set out in educational intelligence for schools.

Where this fits in a joined-up school system

Reducing disadvantage is not a module. It is what happens when planning, teaching, assessment, inclusion and family contact all read from the same evidence, so a pupil's difficulty is noticed early, named honestly, acted on in the next lesson and explained at home in language a parent can use. That is the point of an Educational Intelligence Layer, and it is the only version of this work that survives a busy term.