The call comes on a Friday afternoon. A site manager rings in: the gates are locked on the Leeds job and there’s a notice cable-tied to them from a firm of accountants nobody has heard of. The main contractor — the customer who owes you for two months of groundworks — has gone into administration. The MD asks the only question that matters. What does this cost us?
In most firms I audit, nobody can answer. Not that day. Not because anyone is slow or careless, but because the answer is scattered across six places that have never been added up: the accounts package, the quantity surveyor’s application spreadsheet, the retention list, a drawer of delivery tickets, the hire desk and the order book.
Ten a day
The Insolvency Service publishes the official count. In the twelve months to May 2026, 3,803 construction firms in England and Wales entered insolvency — more than any other industry, roughly one in six of all company failures where the trade is known. That works out at more than ten a day. Insolvency here means the formal procedures: mostly liquidation, where the company is wound up and whatever it owns gets sold off, and sometimes administration, where an outside specialist takes over to rescue what can be rescued. From where you stand the difference barely matters. Money you were owed on Thursday becomes a claim in a queue on Friday.
And the queue has an order. Secured lenders — typically the bank — stand at the front. Certain employee claims and some taxes come next. Ordinary trade creditors — unsecured creditors, in the jargon — stand at the back, and the back of the queue often gets nothing. When ISG, one of the country’s biggest contractors, went into administration in September 2024, the administrators said they did not expect there to be any money at all for the unsecured creditors of most of the group’s companies.
The number nobody can see
Here is what your real exposure to any one customer is made of. Invoices raised and unpaid — the easy part; the ledger has that. Applications for payment submitted but not yet certified — work you’ve claimed for that they haven’t agreed yet; that lives with your QS. Work done since the last application — often recorded nowhere at all. Retentions — the slice held back from every payment as security against defects, usually around five per cent, half released when the job completes and half a year or so later. I’ve written before about how badly retentions get tracked; on an insolvency it’s worse, because retention money is not normally kept in a separate pot. It sits in the customer’s bank account with everything else, and you queue for it with everyone else. Then materials you’ve delivered to their site and not been paid for. Then plant on hire to that job in your name.
Every one of those numbers is knowable. Somebody in your firm knows each of them. The problem is that no one place holds them together, so “what would it cost us if they went under” takes three days and an argument to answer — and it only gets asked after the gates are locked.
The first two days are the expensive ones
Speed matters more than people expect, because the first decisions can’t wait. If your terms of sale include a retention of title clause — wording that says goods stay your property until they’re paid for — you may be able to take back materials sitting on the customer’s site. But only what hasn’t been built in yet, and only if you can prove which materials are yours, and only if you get there before they vanish. There are deliveries booked for next week that need stopping. Plant hired to that job keeps charging by the day until someone rings the hire company. If you carry credit insurance — cover that pays out when a customer fails — the policy will have a deadline for telling the insurer. And your own subcontractors on that job are watching to see what you do.
A firm that can pull the whole number in an hour makes those calls on Friday afternoon. A firm that needs three days makes them the following Wednesday, after the site has been picked over.
What this looks like fixed
This is not an argument for a monitoring platform with a dashboard of credit scores. The fix is smaller and more useful: one view per customer, showing unpaid invoices, uncertified applications, retentions with their release dates, stock on their sites, hire running against their jobs, and how much of your forward workload depends on them. All of it pulled from records you already keep, refreshed without anyone retyping anything.
Once that view exists, quieter benefits fall out of it. You notice that one contractor has drifted to two-fifths of your turnover — a decision you never actually made. You chase retention releases on the date they fall due instead of when someone remembers. You start giving customers internal credit limits, the way the builders’ merchants have always given them to you, and you stop handing the next job to the slowest payer just because the enquiry came in.
Where this doesn’t apply
Be honest about whether you need this. If you run four or five jobs a year for two or three customers you’ve known for a decade, you can hold your exposure in your head, and you should spend the money on something else. If your book-keeper already reconciles by customer every week and the QS’s spreadsheet is kept honest, a whiteboard will do. And no report stops a collapse from happening — if your real problem is that one customer is sixty per cent of your work, that’s a sales problem, and software won’t sell for you.
But if you’re a firm with twenty live jobs across a dozen counterparties, and the honest answer to “what would their failure cost us” is a pause and a guess, the number is worth being able to see. It costs far less to build than one bad Friday. Get in touch.