One project, many contractors, and the gaps between them

A project of any size is built by agencies that do not work for each other. Civil, steel, plumbing, electrical, lifts, facade, finishes — separate contracts, separate supervisors, separate programmes, separate commercial interests. Each reports to you. None reports to another.

Almost every delay on such a project lives in the handover between two of them, and not inside either.

That is an uncomfortable thing to establish, because a delay in a handover has no owner. Ask the civil contractor and the front was ready. Ask the plumbing contractor and it was not. They are describing the same wall on the same day, and in each one's own records both are telling the truth.

Every programme is right and the set is wrong#

Each contractor prepares a programme. Each programme is internally consistent — sensible durations, correct sequence within that trade, resources that add up.

They are prepared separately, and each one contains assumptions about what the others will have done by a given date. Those assumptions are never written, because a contractor's programme is a description of his own kaam and not of yours. Put the programmes side by side and the assumptions contradict each other. Rarely by much. Usually by days, which is exactly the size that nobody escalates and everybody absorbs.

By month four the absorbed days are a month, and the discussion about whose month it is cannot be settled, because no document ever recorded who owed what to whom on which date.

Failures of sequence#

The front that was not ready. The next trade mobilises, arrives, finds the area not handed over, and stands. He has a claim for idling and you have a week gone. Nobody told him to hold, because the person who knew the front was late and the person who called the trade forward are not the same person.

The conduit that was not cast in. Electrical was to lay conduit before the pour. The pour happened. Now it is chased into finished concrete — slower, dearer, structurally unwelcome, and argued about for a month. The cost of the miss is many times the cost of the coordination that would have prevented it.

Two trades in one space on one day. Both were scheduled correctly by their own planners. Neither knew of the other. One wins the space and the other goes home, and which one wins has nothing to do with which is on the critical path.

A hold point nobody was told about. Work may not proceed past a stage until it is inspected — by your engineer, a consultant, an external party. The trade did not know, or knew and did not say, and the work is now covered. It is opened up, or accepted with a note. Both outcomes cost.

Failures of scope and material#

Damage by the following trade. The floor was accepted, then a scaffold went up on it and gouged it. Who pays turns entirely on whether anybody recorded the condition at handover. Without that record it is one man's word against another's, and the man still on site when the defect surfaces is the one who pays.

Material supplied by one, consumed by another. Ordinary on any job — the main contractor's cement used by the plastering agency, steel issued to a fabricator, shuttering lent across trades. If the issue is not recorded against a contract at the moment it leaves the store, it gets reconstructed at final bill from memory, and memory is generous in whichever direction is being asked.

Scope each party believes is the other's. The small item between two packages — the sleeve, the puddle flange, the cutout, the last two metres of a service run. Each contract's wording lets its own reader conclude it is not his. It surfaces when the work is needed, at which point it is priced as an extra by whoever will do it.

The mechanisms that actually help#

None of these is clever. They are all about writing something down before it is contested.

A written interface at each boundary. For every pair of trades that meets: who hands what to whom, in what condition, on what date. The condition carries as much weight as the date — "slab ready" and "slab ready, cleaned, levels checked, cutouts marked" are different obligations, and the difference is a week. This belongs in the contract document rather than a meeting, which is the same argument as what a work order or purchase order has to establish.

Joint inspection at handover, with a record. The two supervisors walk the area together, and the outgoing trade's condition is recorded on the day it stops being his responsibility. Photographs, and a line both sign. This one habit removes most damage disputes, because it converts them from an argument about the past into a comparison of two records.

One dated place where the sequence lives. Not each contractor's programme. One sequence held by you, which every trade reads, and which records when each handover actually happened against when it was due. The daily progress report is where the actual date comes from, if it is written to be read rather than filed.

A route for the boundary problem itself. Interface issues stall because they belong to two people, and a thing that belongs to two people belongs to nobody — the general case of why tasks stall between people. Somebody has to be able to raise it early without it reading as a complaint about a colleague, and escalation without shame is the condition that makes early raising possible at all.

The short version#

Contractors deliver their own scope competently and drop the joints between them. The delay you cannot recover is the one that belonged to two people.

Write the interface down — who hands over what, in what condition, on what date. Inspect at every handover with both sides present and record the condition. Keep one dated sequence that shows when the handover actually happened, not when it was meant to.