Rough-in inspection

Why does an inspector sign off wiring and pipework before the walls go on rather than when the job is finished?

Everywhere else we inspect the finished thing. Nobody signs off a car halfway through assembly, or a meal before it is plated. Yet on a building site the inspector arrives at the least presentable moment — cables draped through joists, pipe ends capped, not a wall board in sight — signs it off, and leaves before the job looks like anything. Then he comes back at the end and looks at rather little.

If inspection is meant to tell us whether the work is any good, why do it when the work is not finished? The instinct is that this is scheduling bureaucracy. Let us see whether something harder is going on.

Start by separating two things that get bundled together as "inspecting". One is the quality of the work. The other is the cost of finding out. The first does not change when the plasterboard goes on. The second changes enormously — and, importantly, it changes suddenly rather than gradually.

Ask what it would take to answer "is that cable clipped at the right intervals and not run through a joist where a nail will find it?" the day before boarding: you look. The same question the day after boarding requires opening the wall. Not a harder look — a different kind of act, one that damages finished work and needs the owner's consent. So the evidence is not merely obscured by completion; it is consumed by it.

Now notice something sharper than "you can't see it". Several trade tests change their meaning across that moment, not just their price. Pressure-test a pipework circuit with every joint visible and a falling gauge tells you there is a leak and a wet joint tells you which one. Do the identical test after the screed goes down and a falling gauge tells you only that somewhere in forty metres of buried pipe there is a leak. Same test, same equipment, same reading — but before covering it localises the fault and after covering it merely detects one. Which would you rather hold in your hand when someone has to pay for the remedy?

The electrical side has the same shape: an insulation resistance reading taken while a run is exposed establishes that the installation left the electrician sound, which matters for liability when a later trade's screw makes it unsound.

Then there is an incentive argument the trades feel more keenly than the physics. The cost of putting a fault right does not rise smoothly with time; it jumps at the moment of covering. Placing the mandatory hold point immediately before that jump puts the deadline exactly where compliance is cheapest — which is why builders resent the waiting far more than the inspection itself.

And the inspector is doing one more thing that is easy to miss: sampling. He cannot follow every cable to its end, so he reads the workmanship where it is visible and treats it as a proxy for the runs he never sees — with the usual weakness of proxies, that it is exactly the inference a dishonest operator would exploit.

Is any of it mere inertia? Partly, and it is worth saying plainly. A staged hold point is also a fee-earning gate and a scheduling chokehold, and regimes keep hold points after the technology that justified them has moved on. But where a genuinely non-destructive later test exists, the pre-cover requirement does tend to soften: buried drainage is often accepted on a camera survey, precisely because there the evidence survives concealment. That pattern is the tell that observability, not ritual, is doing most of the work.

The model is not "inspect early because early is better". It is that verification should happen at the moment when the evidence is cheapest and most informative, and that for concealed work this moment has a hard edge — after it, the information does not degrade, it disappears.

Two refinements follow. First, the final inspection is not a redundant second look; it answers a different question — did anything get damaged, altered or wrongly connected in the fitting-out, and does the finished system behave? Neither inspection can answer the other's question, which is why both survive. Second, the rough-in hold point works only where the interval between the two is short enough that nobody is tempted to cover up and hope. Where the wait is long, sites quietly board over and rely on the inspector's goodwill, and the regime hollows out.

Here is the test that would break the account. If observability is the driver, then work that stays permanently exposed — surface-mounted conduit, services in an open plant room, a barn with no ceiling — should attract no equivalent pre-cover hold, and its inspection should migrate to completion. Find a jurisdiction that demands a rough-in sign-off on a fully exposed, permanently accessible installation that will never be concealed, and the observability story is wrong; procedure would be carrying the load instead.

• What a tradesman prices in when quoting work behind a wall nobody has opened — the same information gap, met from the other side.
• Why post-maintenance testing exists on plant and machinery, and what it can and cannot catch before handover.

Every term the collection defines is gathered in the glossary.