Construction insolvencies are a margin story, and admin is part of the margin
Construction persistently ranks among the sectors with the most insolvencies in the statistics published by the Insolvency Service. The usual explanations are interest rates, material prices and late payment, and they all play a part. But the underlying story is simpler: thin margins leave no room for administrative leakage, and most firms leak more through admin than they realise.
That sounds bleak. It is actually the most hopeful reading available, because you cannot control interest rates, but you can absolutely control whether a variation gets billed.
This post makes one argument: the admin that firms treat as overhead is, in a thin-margin industry, part of the margin itself. Protect it and you become noticeably harder to kill.
Why does construction keep appearing near the top of the insolvency tables?
Partly structure. The industry runs on subcontracted pyramids, long payment cycles, retentions and pricing agreed months before the work is done. Every one of those features moves risk downwards onto the firms least able to carry it.
But mostly arithmetic. Take an illustrative subcontractor turning over £2 million at a net margin of 3 per cent. That is £60,000 of profit for the year. One £15,000 variation done on a verbal instruction and never invoiced removes a quarter of it. One payment application that misses the contractual date pushes a month of cash into the next cycle. One dispute defended with thin records can consume the rest, win or lose.
Firms in this position rarely die of a single dramatic event. They die of accumulated small losses meeting one bad quarter. Which is why the fix is not dramatic either: it is stopping the small losses.
Where does the margin actually leak?
Four leaks account for most of the damage, and none of them looks like a crisis on the day it happens.
| The leak | How it happens | Why nobody notices |
|---|---|---|
| Unbilled variations | Work done on a verbal instruction, never priced or invoiced | Site did the work, so the job feels finished |
| Missed application dates | The payment cycle date slips by a day or two | Cash arrives a month late and gets filed as "timing" |
| Records too thin to defend | The story of the job lives in scattered messages and memory | The cost surfaces a year later as a settlement |
| Retentions never chased | Final accounts close and the retention quietly ages | It is nobody's job to ask |
The first row is the classic. An instruction arrives with the cost left "to be agreed", the work proceeds because the programme demands it, and the paperwork never catches up. The third row is the slow one: disputes are never born big. They start as a small gap in the record that nobody closed, and by the time they surface the evidence you needed was six months ago.
What does the Construction Act already give you?
More than most firms use. The Housing Grants, Construction and Regeneration Act 1996 gives every party to a construction contract the right to stage payments, a payment notice regime with real teeth, the right to suspend for non-payment and the right to adjudicate a dispute at any time.
Notice what all of those have in common: they are date-driven. The payment notice regime rewards the party who applies on time and responds on time, and punishes the one who does not. A payless notice served late is not a discount, it is a default. The Act is, in effect, a margin-protection statute for firms with disciplined diaries, and a trap for firms without them.
So the question is not whether the protections exist. It is whether your application dates, notice deadlines and retention release dates live in a system, or in someone's head.
How does automation defend a margin?
Not by doing anything clever. By making the boring things impossible to forget.
The pattern that works is capture at the moment of the event, then chase without human effort. A variation gets recorded the moment it is instructed, from the same site message that instructed it, so the invoice has a source document. Application dates produce a drafted application before the deadline rather than a reminder after it. Retention release dates sit in a ledger that surfaces them when they fall due. The job record assembles itself day by day, so if a dispute ever does arrive, the evidence already exists.
None of this needs new behaviour from the site team, which is exactly why it works. The firms that struggle are the ones waiting for people to develop paperwork discipline under pressure. The firms that get ahead remove the need for it. The gap between the two is the real cost of doing nothing: not a visible expense, just a margin that never arrives.
What does the resilient firm look like from the outside?
Bankable. That is the positive flip that gets missed when insolvency is discussed as weather rather than as something firms can build against.
Credit insurers, bondsmen and main contractors running due diligence all look at the same things: does this firm invoice promptly, apply on time, know its retention position and hold records that would survive scrutiny? A firm that plugs its admin leaks does not just keep more of its own margin. It gets paid faster, wins the small arguments early enough that they stay small, and looks like a safer counterparty to everyone deciding whether to trade with it.
AI Metric builds this kind of capture-and-chase system for construction SMEs, but the point stands whoever builds it. The insolvency tables are not a forecast of your future. They are a list of leaks, and every one of them can be plugged with tools that already exist.