The Competition and Markets Authority spent two years working out why Britain buys infrastructure badly, and it landed on a true but incomplete answer. Fragmentation, short-termism and a system that rewards process over outcomes are all real. None of them touch the moment where a project's fate is actually sealed: the point at which someone signs off a programme that was never deliverable, and nobody in the room was equipped to say so.
On 8 September, the CMA's chief executive, Sarah Cardell, told a room of MPs at the House of Commons that the UK government spends "around £400 billion buying goods, services and infrastructure" every year, and that the country is "not deploying it as effectively as we could." Her speech named civil engineering directly. "Fragmented accountability, inconsistent procurement approaches and uncertain pipelines can weaken incentives for suppliers to invest in skills, capacity and innovation," she said. She also put a name to something every planner has watched happen from the inside: procurement teams are "judged primarily on procedural compliance, risk management and short-term value," producing, in her words, "a system that rewards process more than outcomes."
That diagnosis lines up with what the CMA found back in May, when its Civil Engineering Market Study put a number on the road and rail sector: £19 billion a year in public spending, held back not by a shortage of contractors but, as Construction News reported it on 9 September, by problems that were "not primarily the result of insufficient suppliers." The market, the CMA said, was "not working as well as it could."
I don't disagree with any of that. I think it's also not where the actual damage happens.
Every fragmentation problem, every short-termism problem, every incentive problem the CMA describes shows up on site as the same artefact: a programme that was accepted, funded and announced before anyone independently tested whether it could be built in the time stated. By the time a scheme is public, has a minister attached to it and a completion date in a press release, the schedule has usually already stopped being a working plan and started being a promise. Reviewing procurement structure after that point is rearranging the furniture in a house whose foundations were never checked.
Here is the pattern, seen from the reviewing end rather than the policy end. A buyer sets a delivery date driven by a funding cycle, a political milestone or a business case that was approved on an earlier, rounder number. A tender programme is built to fit that date, because a programme that admits the date is wrong doesn't win the bid and doesn't get the funding released. Both sides know, at some level, that the date is optimistic. Neither side has an incentive to say so out loud, and the contract mechanisms that exist to test it are used as compliance steps rather than as scrutiny. Under NEC4, a contractor submits a programme for the client's acceptance. The clause exists. In my experience it is treated, most of the time, as a formatting and completeness check: has it got the right activities, the right logic links, the right level of detail, rather than a genuine test of whether the dates in it are achievable given the resources, consents and site access actually available. That is not a flaw in the clause. It is a flaw in how rarely anyone brings independent eyes to what the clause allows them to check.
This is the half of the CMA's own findings that its recommendations don't quite reach. Longer-term funding, a transparent pipeline, better coordination between buying authorities: all of it would help contractors plan further ahead, and none of it forces anyone to independently pressure-test the programme at the point where the commitment is made. You can fix every structural incentive in public procurement and still sign off a schedule nobody stress-tested, because the review the CMA is describing is upstream of the moment I'm talking about. Fair to the CMA: its report does gesture at this. Capability constraints, the ability of buying teams to properly assess what they are being sold, is named as one of its five issues. So it would be wrong to say this has been missed entirely. But naming a capability gap and specifying what actually closing it looks like, on a live tender, before the money moves, are different things. Five structural recommendations don't yet describe that second part, and until something does, the gap the CMA has just spent two years documenting will keep producing programmes that were unbuildable from the day they were approved.
Where I'd push back on my own argument: it isn't only buyers. Contractors bid to win, and a programme that honestly reflects risk and uncertainty is usually a less competitive one than a programme that doesn't. A buyer who gets tougher on deliverability scrutiny without also changing how competitive tenders are scored will just get better-disguised optimism, not more honest programmes. The CMA's own point about incentives cuts both ways. Reform that only tightens the buyer's process, without touching what wins a bid, will move the problem rather than solve it.
None of that changes where the actual test needs to sit. A programme is either realistic or it isn't, and that's checkable well before a spade goes in the ground, whether the reviewer works for the buyer, the contractor, or is brought in independently by either side. The CMA has just given Parliament a very well-evidenced account of why the system produces bad outcomes. It hasn't yet described the one gate, the independent look at whether a schedule can do what it claims, that would actually catch it before the money is committed rather than after the delay is public.
If you're staring at a baseline or a tender programme wondering whether it would survive that kind of look, our free schedule health check gives you a scored, DCMA-aligned read in minutes, no sign-up required to see the result: Free Schedule Health Check. For a full independent assurance review before a baseline is signed off, see the Independent Schedule Assurance Review. And for what an independent reviewer actually looks for beyond the headline DCMA score, see our earlier piece, the schedule review checklist.