What trust leaders cannot see

The finding

Every leader we spoke to described the same gap, in different vocabulary. None could say which of the tools their schools pay for actually work. Subscriptions renew by default. Usage is not tracked. Impact is not reported quarterly, termly, or at all.

This is not a procurement failure. It is a capture failure. The data that would answer the question exists, spread across a management information system, a finance package, a provision map and a dozen supplier portals that do not speak to each other.

Fragmentation presents differently at each scale

• At national scale, with dozens of academies, fragmentation is a coordination problem: holding several improvement priorities at once, where attention to one causes another to slip.
• At regional scale, with headteachers holding their own budgets, it is an absence of central oversight: purchasing decisions are made school by school and never reviewed.
• At system scale, it is interoperability: schools hold enormous quantities of data in products that cannot exchange it, which caps what any AI layer on top can do.

Information arrives faster than it can be acted on

One trust pays an external agency for weekly policy summaries, which a director then filters again before sending a short digest to principals. The filtering is manual and continuous. Long updates get ignored, so brevity is a condition of the update being read at all.

Different updates demand different responses. A statutory change requires an immediate policy and recording system. A landscape shift requires an assessment of existing practice. Primary, secondary and specialist provision rank the same update differently. There is no single process, and the sorting is done in one person's head.

Feedback loops run at the speed of the term

Interventions are decided in termly pupil progress meetings and reviewed the following term. Trust-level oversight sits on top of that: one improvement lead gathering data from heads, feeding upward in generic form.

Asked how quickly they would know if a school was drifting from its objectives, one chief executive was direct: you never quite know what is going on across six sites when you are relying on one person's feedback and data once a term.

Scepticism about AI is an asset, not an obstacle

Leaders were consistently sceptical of AI claims: products marketed as intelligent that are not, selective vendor evidence, peer recommendations that may themselves be vendor-influenced, and evaluations that are not comprehensive enough to rely on.

One trust reported switching a maths intervention from one-to-one human tutoring to an AI equivalent. It cost more and produced worse outcomes. The programme was discontinued.

Where confidence scores, editable outputs and explicit flagging of data gaps were shown, the response was positive in every conversation. Human-in-the-loop is not a hedge in this market. It is the entry requirement.

What follows

The sharpest and most consistent gap is evaluation. Not more data, and not more tools, but a layer that can say which of the existing tools earns its place, and at what cost.

Two data streams would have to be added to make that possible: spoken observation captured as it happens, and longitudinal context that accumulates across a year rather than resetting each term.

Method and anonymisation

Structured discovery calls with executive and national-director level leaders in English multi-academy trusts, collectively responsible for around 50,000 students. Recorded with consent, coded from transcripts, and reported here in anonymised paraphrase. No trust, school or individual is identified.

educave.ai · Free to copy, forward and use inside your institution, with attribution.