article
Why you cannot add up UK contract values
Every published figure is real. Adding them together still produces a number nearly three hundred times too large — and the reason is worth understanding before you trust any table that ranks suppliers by money.
· figures as at the date of writing
Add together every contract value published on the UK’s procurement services and you get £88.1 trillion. British public procurement is worth somewhere around £300 billion a year. The total is out by a factor of nearly three hundred, and every figure feeding it was published exactly as it appears.
One framework, four thousand identical numbers
G-Cloud 15 is the government’s software buying framework. 4,232 suppliers were admitted to it, and the notice records an award to each one. Every single one of those 4,232 awards carries the same value: £14,000,000,000.
That figure is not what any supplier was paid. It is the ceiling of the whole framework — the maximum that everyone on it, together, might spend across its lifetime. The publisher has attached the entire ceiling to each supplier because the notice format has one value field and no way to say this is a shared limit.
Add that framework up supplier by supplier and you get £59.2 trillion from a single contract notice. One framework accounts for 67% of the national total.
How common this is
It happens across the corpus. Of the 34,010 contracting processes with more than one award carrying a value, 21,792 — 64% — give every supplier the identical figure. That is the signature of a ceiling published per supplier rather than a set of prices.
Any league table built by summing values is therefore not measuring who received public money. It is measuring who appears on large frameworks, and how many suppliers those frameworks admitted.
What the sum actually ranks
Rank suppliers by total published value and the top of the table reads:
| “supplier” | summed value | |
|---|---|---|
| 1 | Willmott Dixon Construction | £276bn |
| 2 | A list of successful suppliers can be found in the attachments | £270bn |
| 3 | Dodd Group (Midlands) | £225bn |
Look at the second row. It is not a company. It is a sentence, typed into the supplier field by a buyer who had too many winners to list, and credited by the arithmetic with a quarter of a trillion pounds of public money.
A sentence cannot receive public money. The method cannot tell.
What to use instead
Two things survive contact with this data.
Counts. How many contracts an organisation has won is unambiguous: a supplier either appears on an award or does not. That is how the supplier league table here is ranked, and why the top of it is Softcat, Insight and Phoenix Software — the firms that genuinely win the most public work — rather than whoever sits on the largest framework.
Medians, taken across values not flagged as ceilings. A median describes the typical size of a contract without letting one £14 billion framework speak for a thousand ordinary ones.
There is exactly one page here that does add values up: contracts awarded without competition. A direct award goes to a single supplier with no framework behind it, so the ceiling problem barely arises — and that page still shows how many of each buyer’s awards its total was built from, because 36,681 of 63,649 direct awards carry a usable value and the rest would otherwise vanish silently.
Why this matters beyond arithmetic
Every commercial procurement tool sells a “top suppliers by value” view. So does most journalism using this data. If the underlying figures behave the way they do above, those tables are ranking framework operators and data-entry accidents, and presenting the result as a picture of where public money goes.
The unglamorous version — counts and medians, with ceilings excluded and said so — produces smaller numbers and fewer headlines. It has the advantage of being true.
Figures as at 16 August 2026, from 742,925 notices published on Contracts Finder, Find a Tender, Public Contracts Scotland and Sell2Wales under the Open Government Licence v3.0. The live versions are in the reports. If you think something here is wrong, tell us — that is the most useful message we get.