Play around with AI all you like, but until you stop your margin leaks from normal operations, you’re simply automating loss-making processes, argues YardLink’s Neeral Shah
Almost every conversation in the industry right now runs through the same handful of ideas: autonomous plant, generative design… and AI-powered scheduling.
Don’t get me wrong, these are important conversations. But they are focused on several steps downstream of where the industry is actually losing money.
We manage procurement data across more than 19,000 projects delivered nationwide. The picture our data shows is consistent, and it has nothing to do with a lack of AI-powered machinery. It comes from faulty kit, failed collections and unvetted suppliers, all rippling across every tier of the subcontract chain and well before any AI system gets near the project.
The cost of procurement failure
Let’s start with what procurement failure actually costs. For a mid-sized main contractor turning over £12m, failures across supply chain logistics can account for up to 8% of annual revenue, close to £1m a year. That figure is made up of smaller, less visible losses that compound.
A failed or delayed delivery costs somewhere between £750 and £800 once labour and programme impact are counted, escalating to £15,000 in lost productivity per site over a project lifecycle, or £75,000 across a five-site operation.
Separately, procurement teams lose up to a full working day a week to sourcing and procurement-related admin, worth £25,000 to £35,000 in recoverable capacity a year, before finance teams even begin reconciling the invoice queries that follow. None of this appears as a clean line item. All of it erodes a margin the business thought it had secured at tender stage.
“A sector that layers advanced automation on top of an unreliable, poorly vetted supply chain will simply automate its existing losses faster and at greater scale.”
Non-compliant documentation
Paperwork compounds the exposure rather than sitting apart from it. We’ve found that around a third of waste carrier documentation across UK construction supply chains is non-compliant or missing critical regulatory information.
That is not a filing problem. Duty of care for waste does not transfer when the lorry leaves site – it stays with the contractor who produced the material. A missing carrier registration or an incomplete waste transfer note is a live compliance risk that can trigger a penalty or halt a project long after the cheapest supplier was appointed and the paperwork was signed off without a second look.
The difference between managing this and not managing it is measurable, and it is larger than most procurement teams assume. Where supplier vetting and delivery documentation are checked systematically, structured vetting, automated checks, verified credentials, we see downtime fall by roughly 65% and invoice queries held below 1.5%.
Where that discipline is absent, both figures move sharply in the other direction, and the gap widens with every additional site added to a live portfolio.
Commercial and reputational risk
That gap does not sit evenly across the chain. Main contractors carry the commercial and reputational risk when a site stalls, but they rarely make the supplier vetting decisions that cause the stall. Those calls are often made several tiers down, under time pressure, by whoever is available that week.
Subcontractors absorb the programme delay when equipment fails to arrive, without the leverage to demand better terms from a supplier they did not choose. Suppliers, in turn, are frequently judged on day rate alone, which rewards exactly the behaviour, cutting corners on vetting and maintenance, that produces the failures in the first place. Across a portfolio of live sites, that misalignment scales faster than any single project’s budget can absorb.
None of this is an argument against AI in construction. Autonomous plant and generative design will earn their place over the next decade. But a sector that layers advanced automation on top of an unreliable, poorly vetted supply chain will simply automate its existing losses faster and at greater scale. The industry does not have a technology problem. It has a reliability problem sitting upstream of every tool built to solve it.
The contractors who will get genuine value from construction’s AI wave are the ones fixing this now: quantifying failed deliveries, formalising supplier vetting, and making compliance a condition of appointment rather than an afterthought. Fix the upstream problem, and every pound spent on innovation from here goes further. Leave it alone, and the industry will keep investing in the wrong end of the chain.
