Signed on trust

A bid worth tens of millions rests on about a hundred files. The meeting where it is signed off takes 45 minutes. Actually reading everything would take four people a week. What gets signed is not the documents. It is trust.

A carpenter's pencil and a steel try square on a concrete ledge

The signature certifies trust, not a reading

The scene repeats in contractor offices every tender season. A bid worth tens of millions of kronor is due. Underneath it sits the tender package: around a hundred files of administrative conditions, technical descriptions, drawings, quantity schedules and appendices, with the contractor's own calculation stacked on top. Before the bid goes out, the people responsible gather to walk through it and sign. The meeting takes about 45 minutes. Half the people at the table have other bids in motion the same week.

Reading those hundred files properly, with the attention it takes to notice the one line that decides the outcome, would take four people about a week. The meeting has three quarters of an hour. So the meeting cannot review the documents; it reviews the people who handled them. Did the estimator seem confident? Has this client's paperwork behaved before? The signature at the end certifies a chain of trust, because it cannot certify the reading.

The meeting and the reading are 200 to 1 apart

Roughly two hundred to one. The figures below come from Yesper's conversations with Nordic contractors, rounded on purpose and stripped of anything that would point to a firm or a procurement. The shape is the point, not the decimals, and the shape repeats wherever we ask. None of the figures are exact, but everyone we show them to recognizes them.

Item Approximate figure
The bid Tens of millions of kronor
The files underneath it About 100
The sign-off meeting About 45 minutes
Reading every file once Four people, one week
Gap between meeting and reading Roughly 200 to 1

Nothing in the table is a scandal. It is a budget. Forty-five minutes is what a senior team can spare for one bid among several running at the same time, and four person-weeks of reading is hard to justify for a tender the firm may well not win. Every hour spent reading a losing bid is an hour billed to nobody. The triage is rational for every single bid, and risky for the sum of them.

Half the cost gets the least scrutiny

The same arithmetic, one tier down, decides an even larger number. In a typical contractor's calculation, roughly half of the project cost is not the firm's own work at all. It arrives as quotes from subcontractors, and it arrives late: in a pile, during the final days of the bid window, because the subcontractors are being squeezed by the same deadline one level below. There is rarely time to question or verify them; they go into the calculation largely as they landed. Many carry their own reservations, each of which would need to be checked against the tender documents.

Those quotes get judged on gut feel. Does the number sit where the last few similar jobs sat? Is anything obviously missing from the scope? Gut feel of that kind is a real skill, decades of priced projects compressed into fast pattern recognition, and it is usually right. But the plain version deserves to be written out: the largest single block of cost in the calculation gets the least tool-supported decision in the entire bid.

One contract can run to 20,000 pages

More than any team will ever do. One maintenance contract, a single contract rather than a portfolio, ran to more than 400 files and close to 20,000 pages. A field technician who was shown the archive said what everyone in the industry quietly knows:

Ingen skulle någonsin läsa det här — det är ett arkiv för maskiner, inte människor.

A field technician at a rail maintenance contractor, from Yesper's customer conversations. "No one was ever going to read this: it is an archive for machines, not people."

He meant it as a description, not a complaint. Twenty thousand pages is not a document set sized for human reading; it is decades of requirements, standards and liability clauses, each layer added for a reason, none ever removed. Document volume kept growing. The hours in a bid week did not. Every individual page made sense to someone; the whole no longer makes sense to anyone.

This is time arithmetic, not negligence

No, and that is worth saying clearly. The person signing is doing the only rational thing available: putting the scarce reading hours where experience says they pay, and covering the rest with trust in colleagues who have earned it. The failure is not in the room. It happened slowly, upstream, as the paperwork around construction grew past what any realistic staffing could read back. Every tier of the chain runs the same quiet triage, from client to contractor to subcontractor.

Trust deserves better material to work with, though. When reading a hundred files no longer costs four person-weeks, the 45-minute meeting does not get shorter; it gets better, because its minutes go to judgment instead of hope. Then the signature returns to meaning both things at once: I trust these people, and someone has actually read the files.

Benjamin Glaser Co-founder at Yesper. Writes about AI and the industry that builds the world. benjamin@yesper.ai

Yesper is the AI civil engineer for construction and infrastructure. AFRY, COWI, NRC Group and other Nordic firms use it to halve the time on a study, rerun calculations in minutes, and catch errors that would otherwise slip through. Get in touch if you'd like to see what it can do for you.

Book demo