Newham Council needed £32m of savings as it entered 2025-26. It applied for exceptional financial support, and secured government permission to raise council tax above the legal cap of 4.99% to 8.99%.
Meanwhile, its Microsoft bill increased by 80.5% over three years, from £2.13m to £3.85m. Asked whether it has ever run a competitive procurement in which Microsoft could have lost, the council had no process to show.
Newham’s failure to run competitive tendering between strategic cloud and software suppliers hits starkly against a background of financial distress, but it is not unusual.
Computer Weekly analysed Freedom of Information (FoI) responses from 37 councils – provided by a source that requested anonymity – and found 32 had never run a competitive procurement in which an alternative platform could have won. Competition only took place between resellers of the identical Microsoft product, not between Microsoft and a competitive platform.
In this article, we outline the nature and extent of that pattern across a whole tier of government, and find that for up to £400m of budget spent, Microsoft was the default answer.
That all occurs as the Competition and Markets Authority (CMA) decides whether to designate Microsoft with strategic market status (SMS) under the Digital Markets, Competition and Consumers Act.
SMS status would bring powers to impose binding conduct requirements, mandate interoperability and restructure commercial arrangements across a software estate that touches an estimated 20 to 30 million organisational users in the UK.
The money, and what sits alongside it
The 29 councils that disclosed a usable 2025-26 figure spent £62,399,176 between them on Microsoft products – an average of £2.15m each. The biggest disclosed spend in the sample is Birmingham’s £4.32m, a figure it gave as a combined total with no year-by-year split. The biggest reported increase belongs to Newham (+£1.72m) and Lancashire (+£1.25m), which together account for 69% of the sample’s total three-year rise.
Scaled against core spending power – the government’s own measure of the funding a council has for its services – that extrapolates to a bill for English councils of roughly £281m a year, with a plausible range of £241m to £400m.
It is tempting to see a relationship between that spending and the financial distress that runs through the councils sampled. However, across the 11 councils that published both a single-year budget gap and comparable three-year spending, there is no relationship between the size of the gap and the direction of the Microsoft bill.
The two councils with the highest tax rises in the sample cut their Microsoft spending: Bradford 6.6% against a 9.99% rise, and Worcestershire 1.4% against 8.99%. What is true, and provable, is narrower and harder to dismiss. As many as nine of those 11 increased their Microsoft spending while carrying a published gap, and 10 of the 11 have never tested the platform.
How the numbers were produced
The evidence base comprises 37 responses to FoI requests that asked 16 questions. These covered five broad areas designed to scrutinise local authority expenditure and governance. They begin by examining tendering history and procurement frameworks for productivity software before diving into detailed current spending, Enterprise Agreement expiry dates, and adoption timelines for specific cloud and AI tools like Microsoft Copilot.
The inquiry further evaluates internal oversight by asking about value-for-money assessments, supplier addition policies, cost benchmarking and contract publication. Finally, it investigates high-level engagement between IT leadership and Microsoft, the specifics of secured social value commitments, and broader IT budgets alongside cashable transformation savings.
Several of the responses arrived as image-only scans, so Computer Weekly OCRed them to make them machine-readable.
Also, comparisons need a common denominator, because a £30m hole in a London borough is not the same thing as a £30m hole in Bolton. So, budget gaps have been calculated as a share of core spending power, and the national figure scaled on the same basis. The sample skews towards larger, urban authorities and contains no shire districts, which is precisely why a mean-per-council extrapolation would overstate the national number.
The machine that produces this
The reason 32 councils can behave identically without coordinating is that they share a buying mechanism. Namely, 26 of the 37 buy through a Crown Commercial Service framework, and seven through KCS Procurement Services, a trading arm of the Kent County Council-owned Commercial Services Group. That means 33 of 37 councils are funnelled through two buying bodies. The single most common route is CCS’s RM6098/Technology Products & Associated Services 2 (TePAS 2) Lot 3.
Those frameworks compete at the level of the reseller, not the platform. The product and its underlying price are identical whichever reseller wins – a mechanism Crown Commercial Service has repeatedly reshaped without changing. And the reseller market is itself concentrated: Phoenix Software supplies 12 of the 37 councils, while the formerly separate – now merged – Trustmarque Group and Ultima Business Solutions account for 21 of the 37 councils that responded.
The councils explained the outcome in their own words.
Salford wrote: “The council always go to market via Procurement and Crown Commercial Services (CCS) for our Microsoft licensing. These are open tenders, allowing suppliers within the relevant lots to bid.” That is reseller competition described as open tendering.
West Northamptonshire explained why nobody pushed further. It said replacing “all or part of the Microsoft ecosystem” would “require massive investment in time and funds for re-platforming, re-training staff and re-architecting security protocols”, in an application environment that spans social care, revenues, benefits and housing repairs.
Nobody checks, and nobody says why not
The tender question and the benchmarking question tell the same story from different angles. Only one council in 37 – Worcestershire – benchmarked Microsoft’s licensing costs against an alternative in the past five years, while 24 did no benchmarking at all. Five had compared resellers, buying routes or peer councils, which is not the same thing. Liverpool’s November 2025 exercise, for example, compared two ways of purchasing the same Microsoft products.
Answers to the value-for-money question were revealing:
- Gateshead: “No, assessments are based on technological merit.”
- Worcestershire: “Yes, but not retained as records.”
- Hillingdon: “No – no VFM assessments held for Microsoft services.”
- Newham and Camden: “Information not held.”
- Hammersmith & Fulham: “Yes – details immediately easily traceable; may take a disproportionate amount of time to collate.”
Camden’s full answer to the tender question shows it held no record of whether it ever tested its biggest software relationship, and no record of why not. It said: “Information not held. The Council does not hold information confirming whether an open market tender of the type described has previously been undertaken, nor recorded information setting out reasons why such a procurement has not been undertaken.”
Copilot on the nod
A key case is that of artificial intelligence (AI) adoption, where 31 of the 37 councils asked have adopted or are piloting Microsoft Copilot. That’s a conservative figure since three of the remaining six withheld the price as commercially sensitive, which implies they hold the product.
Some 20 disclosed a figure, totalling £1,185,308 a year; Lancashire alone is £339,000 of that, nearly 29% of the disclosed total. Of the 31, only one – Hounslow – says it holds a value-for-money assessment covering Copilot. Merton is the only other council to mention AI, and its “AI solutions exercise currently underway” is happening after adoption.
Of the councils that stated a date, two adopted in 2023, 10 in 2024, six in 2025.
Respondents to the CMA’s investigation, published by Computer Weekly in June, explained the mechanism. That is, that Copilot Chat is bundled free with most enterprise licences in a way that presents Copilot as the default AI tool at work, and channels customers towards full paid M365 Copilot licences.
One respondent said then that bundling, “may seed usage of Microsoft’s AI tools before enterprise customers consider competing enterprise AI solutions” while third-party AI providers face “inferior API, connector or add-on access”.
Microsoft’s position, set out at length in its CMA submission, was that the market is vigorously competitive and that “AI is dramatically lowering the barriers to building high-performing enterprise software”. It pointed to Google Workspace, macOS, Linux, PostgreSQL and Okta as evidence that each layer of the stack is contestable. The councils’ responses to the FoI requests here test that claim from the buyer’s side, and don’t seem to confirm it.
Nothing for the millions
Asked what social value they secured from Microsoft at their most recent renewal, 11 councils recorded none at all, and 19 said the commitment rested with their reseller or their framework rather than with them. Doncaster’s answer was: “Zero – no social value commitment to the council.”
Norfolk’s tender put it plainly: “The Council is not seeking social value from this procurement because the main requirement is for the provision of software licenses on a pass-through basis from a reseller.”
Five councils identify commitments of their own. The best is Croydon’s: £7,500 of community donations, £271 of volunteering time, two apprenticeships a year, skills workshops and three refurbished laptops – £8,042 in total, against the council’s own £2.4m Microsoft bill. That is 0.335% of its spending.
The fair reading, though, was not that no value exists. Gateshead supplied its reseller’s own commitment document – a Phoenix Software statement covering the Microsoft Partner Pledge, Living Wage accreditation, apprenticeships and modern-slavery measures – the only council to do so. Islington’s contract stated the same arrangement: “The contract is with Phoenix so the SV commitments will be by them.”
Social value commitments were overwhelmingly offloaded to the reseller, which means the finding is not that social value is absent, but that councils cannot show what their own spending bought. Given one reseller supplies a third of the sample, that is a significant gap between a council’s duty and its ability to evidence it.
What testing looks like when someone does it
Hackney was the counter-example, and it was a telling one because the council did not set out to prove Google cheaper. It commissioned consultants to cost a move to Microsoft.
The review modelled six Microsoft options against Google Workspace’s £968,437 a year for 6,001 users: from £2,023,537 (2.1× more expensive) for core E3 productivity to £5,242,474 (5.4×) for M5 with Copilot. Its own likely option, E3 plus a security uplift at £2,556,426 (2.6×), would have needed £1m to £1.6m a year of new revenue budget.
Hackney renewed Google competitively, having explicitly rejected a direct award because it “lacks transparency and fails to ensure value for money”, and secured an apprentice for every £2m of contract value.
It did move its legal services to Microsoft for court and police interoperability, while its review credited Microsoft with real advantages that “come at a significantly increased annual cost”. The point is not that Google is better. It is that Hackney measured the trade-off, and 32 of the other 36 did not.
What needs to change?
Spending on Microsoft increased, the AI layer was just bolted on, and tests did not happen. Meanwhile, the councils that did do any comparisons were the ones under the greatest pressure: Hillingdon carried the largest gap relative to its funding in the sample and balanced it only with £150m of exceptional support; Worcestershire needed support to avoid a Section 114 notice.
Three things would change this situation, and all are structural rather than cultural. Frameworks could make platform-level testing a condition of use, rather than a choice no council makes.
Councils could be required to record a value assessment before adding a service from an incumbent supplier – Camden’s answer shows why recording matters, because an undocumented decision is an unknown one. And the CMA’s Strategic Market Status work could reach into public sector licensing explicitly, rather than leaving councils to discover the question on their own.
Until then, hundreds of millions of pounds a year goes on being spent, and almost no councils are asking whether it should be.
