That £3.6bn of school contracts is really £412m

Corrected 25 September 2026. The first version of this piece said the honest total was £2.4bn. That was still wrong, by a long way: it counted framework ceilings, and one £1.05bn contract, as if they were money being spent. The same notices come to £412m.

We found both problems the way you'd expect. The first time, the total looked too big, so we went and checked. The second time, a trust whose schools take in £12.7m a year showed £833m of contracts on its page.

Find a Tender publishes one notice per award, and that notice lists every supplier admitted to it. Our importer stores one row per award per supplier, because that's the only way to answer the question people actually ask, which is who has won work with this trust. The side effect is that summing the amount column counts a multi-supplier award once for every supplier on it.

Embark Multi Academy Trust published an estates works and capital projects dynamic market in March 2026 admitting 116 suppliers. Hornchurch Academy Trust's construction and building works framework admitted 50 to one lot valued at £138.3m. Harris Federation's staffing framework admitted 48. Sum those rows naively and Hornchurch's £138.3m lands in your total fifty times over.

Deduplicate to the award and £3.6bn becomes £2.4bn. That part still stands. A third of the headline was one number wearing several hats.

What we got wrong was the next step. We wrote that a framework's value is a ceiling over its lifetime, and that nobody could tell those apart from real contracts. Find a Tender does tell them apart: every notice says whether it sets up a framework agreement. We just hadn't been keeping that field.

We were wrong about that.

Forty notices on file are frameworks, carrying £906m of ceilings between them. Hornchurch's two frameworks account for £830m on their own, and they're open to any public body in the UK. They're a limit on what everyone together might buy through them, not a forecast of what one trust will spend.

Then there's Red Kite Learning Trust's £1.05bn contract with Pagabo, a company that runs frameworks for other buyers. The value is what Pagabo expects to put through those frameworks over five years, roughly fourteen times the trust's own annual income. The notice doesn't call it a framework, so we now also set aside any award worth more than twice its buyer's annual income. In our data nothing real comes close: no award sits between one and five times its buyer's income.

Our position hasn't changed, it's just firmer: treat any figure for school or trust contract spend as unusable until whoever quoted it tells you how they handled frameworks. It's the easiest mistake to make with this dataset, it only ever inflates, and we made it ourselves.

So £412m is the honest reading of the contracts schools and trusts awarded between February 2025 and the start of September 2026. It's still award value rather than spend, and still only what reached Find a Tender. Our trust, school and supplier pages now list framework awards with the ceiling marked, and leave them out of every total.

Method. Award-stage notices on the Find a Tender OCDS API published 25 February 2025 to 1 September 2026, matched to trusts by Companies House number and then by normalised name. One value per distinct award: the maximum across its supplier rows, which repeat it. Left out of the total: awards on notices flagged as framework agreements (tender.techniques.hasFrameworkAgreement), awards worth more than twice the buyer's latest annual income from its DfE finance returns, and lots whose awards all repeat one value. We first counted 909 distinct awards on 8 September. Buyers have since amended and republished 23 of them, which moved their publication dates past 1 September and out of this window, leaving 886. None of the 23 carried a value, so the totals are unchanged. We now keep each award's first publication date, so an amendment can't move it again.