Over the last 10 years I have read most of the edtech ‘reports’ commissioned by the DfE. The latest, Assessment of the education technology market in England, was published in June. Commissioned from PUBLIC, a govtech consultancy founded in 2017 by Daniel Korski (David Cameron’s former Deputy Head of Policy) and Alexander de Carvalho (Heineken heir and venture investor). The authors were Martina Bolibruchova, Johnny Hugill, David Rayman and Eden Yitbarek, with data from The Data City (a “real-time industrial classification” outfit, i.e. commercial web-scraping) and EdTech Impact, a commercial reviews and market-intelligence platform whose paying customers are edtech vendors.
So where is the bullshit? A good starting point is the claim that there are 1,123 edtech companies with £6.5bn p.a. turnover. That’s total crap and the authors, the DfE and anyone who is serious about this stuff ought to know that. Now before The Data City’s PR team gets excited, this isn’t “SIC codes good, web-scraping bad”. SIC is useless for edtech, there isn’t a code for it, which is exactly why the 2022 report and this one both ended up scraping websites. But let’s be honest about the tradeoffs. SIC (Standard Industrial Classification) is an old broken system, the errors are well known, the Office of National Statistics (ONS) can apportion activity through actual surveys, and when it’s wrong we know roughly how and why. The Data City’s RTIC (Real-Time Industrial Classifications) are a proprietary black box. No published error rates, no way for anyone outside the company to check the list or re-run the numbers. Financials that are half Companies House filing and half algorithmic guesswork. It’s a fantastic tool if you’re a VC hunting for deals or a council that wants to look innovative, that’s what it’s built for and what it’s sold for. It is not a measurement instrument. The DfE used it as one anyway, then published the reading to four significant figures. And their own Appendix 5 lets the cat out of the bag- large companies drive 65% of sector employment, so the “1,123-company thriving ecosystem” is a handful of big firms plus a long tail of minnows whose turnover a model made up. Is Pearson in the 1,123? Nobody knows, the list isn’t published. If they’re in, one company’s global turnover swamps the total. If they’re out, the boundary is arbitrary bollocks. Either way somebody at the DfE should have to answer for it.
Even worse in the 2022 Frontier report (Appendix 1) claimed edtech companies had 41,000 employees and £3.7-£4bn GVA, just for England. This new report quietly recalibrates that 2022 baseline down to 31,300 employees and £1.87bn GVA, halving the previous DfE-published figure, again buried in Appendix 6.
And don’t expect that to change. The Economic Crime and Corporate Transparency Act 2023 was supposed to fix this by compelling even small companies to file a P&L, but implementation keeps being delayed and is currently signalled for April 2028 at the earliest and even then companies may be able to opt out of publication of their P&L. Companies House, having struggled just to get director identity verification off the ground, is taking zero enforcement action against small companies who come and go like mayflies, even when complaints are made. Then the small companies exemption threshold jumped in April 2025 to include companies who meet two of three thresholds; turnover ≤£15m, balance sheet ≤£7.5m, ≤50 employees That’s why someone like CenturyTech, despite its founder’s claims of AI edtech world leadership has always been able to file Small Company accounts.
So if the turnover data mostly doesn’t exist, where did £6.5bn come from? Only nerds like me read Appendix 5, which admits that turnover for non-disclosing companies was “estimated using predictive modelling” and headcount via a K-Nearest-Neighbours algorithm, one model for guessing revenue and another guessing employees. Applied across such broad (read bad) data, this is exactly how you get bullshit headlines quoted to four significant figures such as £6.517m turnover and 8.8% growth. There are no confidence intervals anywhere on the turnover figures, and the employee numbers range from 29,660 to 39,100, a 32% spread! It’s the old story: shit data in, shit data out.
The bottom line is that turnover, no matter how badly calculated, is not market spend, and PUBLIC, their staff, the DfE and anyone with any commercial nous knows that. It’s probably the oldest inflation trick in sector-advocacy and PR, doing exactly what it always does, in this case making edtech look far bigger and more important than it really is to the DfE, whose budgets fund almost all of what schools actually spend.
It’s therefore unsurprising to find, buried a few pages in, the Gross Value Added (GVA) estimate, the actual measure of economic value added, at just £1.8-£2.4bn, just 27% to 37% of the headline figure. Two comparisons put that in perspective. Against the government’s own prior GVA assessment: the DfE’s 2022 market report (see Appendix 1) claimed £3.7-£4bn of GVA for England alone (this report says its £5.93bn or 91% for England which is 6x the governments own demand side estimate). Against actual demand: the government’s own (uncited) estimate of what UK schools spend on edtech each year is around £900m, a Department for Business and Trade figure, while the DfE’s 2019 strategy put spend on learning software and hardware at £470m. Nothing in any serious data supports a £6.5bn, except the egregious turnover-aggregation method PUBLIC used.
This is not a one-off. I remember writing about how bad the DfE’s 2022 Ecorys report, Future opportunities for education technology in England, was (see Appendix 1). You could apply my general criticism of that to pretty much every DfE report since, particularly the November 2022 analysis, The Education Technology Market in England, by Frontier Economics and IFF Research (see Appendix 1). Why? Well the first has more holes than Swiss cheese and the second is where some of the bullshit in this new report comes from.
A key problem is the 2025 headline switches its fundamental geographic base from England to the entire UK. But their own Table 4 gives England-level 2025 data: 27,012 employees. Against the recalibrated 2022 England baseline of 31,300, their own like-for-like number, calculated using their own methodology gives a circa 14% fall in English edtech employment over 36 months! Something that’s never mentioned, anywhere, in a document whose framing is that, “the EdTech sector has grown and matured since the 2022 baseline.”
I could keep going on about how the demand-side claims, like “35% of schools use Arbor“, come from EdTech Impact data on schools within a single MAT (the usage dataset is 276,890 records from a single large MAT, between Feb. 2025 to Feb. 2026). This is ridiculous (and I love Arbor, so it’s no reflection on them): MATs centralise procurement, so all this tells you about is one procurement team’s choices, not the entire English school system. What makes it inexcusable rather than merely lazy is that far better demand-side data exists and the DfE knows it. Teacher Tapp’s Brand Tracker polls a panel of over 10,000 teachers in more than 4,000 schools, re-weighted daily against the School Workforce Census, covering 335 education brands in 2025 with each brand polled by around 3,000 teachers on awareness, usage and recommendation. It won’t tell you exactly what schools spend, nothing will (based on current filing rules not requiring a P&L and/or audit), but for who is actually using what in classrooms it is nationally representative in a way that EdTech Impact’s data simply isn’t. The DfE has even used Teacher Tapp before: the Ecorys 2022 report ran its teacher survey through it but this time the PUBLIC chose to use a vendor-funded review platform’s single-MAT dataset instead.
The icing on this steaming pile of bovine ordure is the four anonymous people who make up the “Lived Experience Board“, one each from edtech procurement, Local Authority administration, an edtech consultancy and someone from “AI”. There are no names, no disclosure of affiliations, no conflict-of-interest declarations, no detail on whether they were paid. We have absolutely no idea whether any is, or represents, an entity that could be defined as (to use legal parlance) ‘a materially interested party’. These four did whatever it was they did across a series of workshops between September 2025 and January 2026.
I believe in transparency, particularly as this report, and every other one in Appendix 1, was commissioned by the DfE and paid for by UK taxpayers (happily I am no longer one, but that’s another story). An unrepresentative, anonymous panel of four is not a mistake, it’s a fundamental governance and transparency failure that falls well short of the Nolan Principles’ standards of openness, integrity and accountability. As the Committee on Standards in Public Life has stated for years these extend to private organisations delivering publicly-funded services, which squarely covers PUBLIC’s work on this DfE contract. Compare the DfE’s own EdTech Evidence Board: named members, public constitution – exactly same department, the same policy area, but an totally opposite transparency regime.
Which brings us to the obvious question: why PUBLIC, and who are PUBLIC Group International? They have no substantive research track record in edtech. At least most of the DfE’s commissioned 2022 reports, for all their faults, were done by experienced education researchers or in partnership with an actual economics consultancy. PUBLIC began life as an accelerator, taking a 3% equity share from startups it helped to pursue government business. In its launch week, it listed serving special advisers as available to advise companies on its paid programmes, names it removed after BuzzFeed News, which characterised the scheme as offering “paid access”, made inquiries about Cabinet Office clearance. By its own account the firm, “started life as an investor and accelerator, supporting and building startups to win government contracts” and, “now delivers those very same contracts”. If that’s not a structural conflict defined, I don’t know what is – a company that built its brand growing the govtech market, accelerating vendors through GovStart, co-hosting the GovTech Summit (with Founders Forum), and now selling delivery and strategy to departments, is then paid by the DfE to tell it how big the edtech market is. Unless you had never dealt with Whitehall and the vast ecosystem that exists around it, you might believe a bullshit number like £6.5bn. Happily, I can debunk that for you.
As for who runs and owns PUBLIC (legally PUBLIC Group International Ltd, formerly Public.io), it is now a subsidiary of Finland’s Solita Oy, itself backed by Apax Partners’ digital growth fund, a tech-focused private equity fund. The deal closed on 31 January 2025, when all 62 pre-existing shareholdings (including Founders Forum) were transferred, leaving Solita Oy (Finnish Business ID 2905040-7) holding 100% of the 3,069,864 ordinary and 820,196 preferred shares. Given how influential PUBLIC seem to be, it’s ironic that they too file small-company “filleted accounts” (audited at least) though the Disclosure Note went to Companies House with the template placeholders still in it, literally reading “(enter name of group financial statements)” LOL (see Appendix 2). A digital transformation partner, transforming digitally. Those accounts show accumulated losses of £4.9m in 2025, improved from a restated £5.4m in 2024. One tiny factoid that may ruffle a few feathers within Labour: among PUBLIC’s cashed-out investors were 8VC Fund II and 8VC Entrepreneurs Fund II, Joe Lonsdale’s US funds (he’s a Palantir co-founder). Silicon Valley money in a firm advising Whitehall. So this is a bad report from what is nominally a UK entity, owned by a Finnish company backed by private equity, with previous Silicon Valley investors. Was any of this disclosed (it’s all public record, because I found it) before the contract was awarded? Hopefully, but I can’t be bothered to keep digging as FOIs are a huge waste of time.
The bottom line is no-one should rely on this tendentious piffle. Instead, read a few books like Class Clowns (Jonathan A Knee) and The Signs Were There (Tim Steer); both several years old. These will give anyone a basic understanding of how the education industry operates and the ability to read and understand basic accounts. Next start reading the company accounts for a few hundred education businesses and you will then have the beginnings of the skills to discern the wheat from the chaff in edtech and beyond.
Some Actions Points
- Publish the 1,123-company list
- The DataCity publish RTIC error rates
- Name the Lived Experience Board and their interests
- Disclose the contract value and pre-award conflict declarations
- Revise using Teacher Tapp for demand-side data.
Appendix 1 working research table, readers are invited to help.
DfE-commissioned edtech research, 2021–2026
| Report | Pub. | Organisation(s) | Named people | Cost |
| Education Technology (EdTech) Survey 2020–21 | May 2021 | CooperGibson Research | Not individually credited | Unpublished — DPS call-off; check Contracts Finder/FOI |
| Future opportunities for education technology in England | Jun 2022 | Ecorys UK (app. Dec 2021; work Dec 2021–Mar 2022, incl. Teacher Tapp survey) | Not individually credited | Unpublished |
| Implementation of education technology in schools and colleges (“What Works in EdTech” Stage 2) | Sep 2022 | CooperGibson Research | Not individually credited | £58,650 — awarded Dec 2021 via dynamic purchasing system call-off, ref con_12998 |
| The education technology market in England (the “2022 report”) | Nov 2022 | Frontier Economics + IFF Research, with a Glass.ai web-crawl for market identification | Jane Aston, Elizabeth Davies (IFF), Maria Guijon, Katharine Lauderdale, Danail Popov (Frontier Economics) | Unpublished |
| EdTech Demonstrator Programme Phase 2 evaluation | Nov 2022 | ImpactEd Evaluation (partnered with DfE from 2021) | Not individually credited | Unpublished — tender ref DFERPPU 20-21/050 on Contracts Finder |
| Education technology for remote teaching | Nov 2022 | Likely CooperGibson (they ran the adjacent EdTech suite) — unconfirmed, verify | — | Unpublished |
| Technology in Schools Survey 2022–23 (TiSS wave 1) | Nov 2023 | IFF Research — commissioned late 2022 for a five-wave biennial survey series | Not individually credited | Unpublished — multi-wave contract, so the award value will be substantial; worth pulling |
| Generative AI in education: educator and expert views | Jan 2024 | HM Government’s Open Innovation Team (OIT), commissioned by DfE’s Digital Strategy Division | Not individually credited | OIT is a cross-government unit that departments pay for projects — cost via FOI |
| Use cases for generative AI in education: user research report | Aug 2024 | Faculty AI (project leadership, tool development and testing), with DfE | Not individually credited | Unpublished — the Faculty contract value is a genuinely interesting FOI target |
| Ofsted study of AI ‘early adopter’ schools and colleges | 2025 | Ofsted, funded by DfE — 21 interviews with schools, FE colleges and MAT leaders | — | Inter-departmental funding; FOI |
| Technology in Schools Survey 2024–25 (TiSS wave 2) | Nov 2025 | IFF Research (same five-wave contract) | — | As above |
| Assessment of the education technology market in England | Jun 2026 | PUBLIC (PUBLIC Group International Ltd), with The Data City (market data) and EdTech Impact (usage data) | Martina Bolibruchova, Johnny Hugill, David Rayman, Eden Yitbarek | Unpublished — priority Contracts Finder/FOI target |
Appendix 2 – why PUBLIC’s 2025 “filleted accounts” make me LOL
Leave a Reply