Due diligence on an outsourced detailing partner

Most due diligence on a detailing supplier measures things that do not predict the outcome. A software list tells you what they have bought. A headcount tells you what they can absorb, not what comes back. References are selected by the supplier, and a reference who had a bad experience is not on the list.

The things that actually predict rework are process questions, and they are uncomfortable to ask only because the answers are so easy to check.

The six questions

1. "Show me a model you produced, and let me open it"

Not a rendering, not a screenshot. The file. Open it yourself and check four things: that the origin and units are what the issue sheet says, that the grids and levels agree with the reference model it was built against, that the elements carry the properties the next stage needs, and that the contents match what the transmittal claims was issued.

A supplier who cannot produce a model to open — for confidentiality reasons, which is a legitimate answer — should be able to produce one they built for themselves.

2. "What do you do when a drawing does not answer a question?"

The answer you want is a process: the question is raised, numbered, and answered against its number, and the answer goes into the model. The answer you do not want is any version of "we use our experience", which is a description of an assumption being taken quietly.

Quiet assumptions are where detailing rework comes from. Not from bad drafting — from a bolt grade nobody confirmed, applied consistently, forty times, arriving as material that has already been cut.

3. "What happens when the work cannot wait for the answer?"

Sometimes it genuinely cannot. The question is whether the assumption gets printed. A supplier who states the assumption on the drawing or the calculation, where a reviewer cannot miss it, has given you a chance to catch it. A supplier who holds it in their head has not.

4. "Who does the connection design, and is it in the same model?"

Splitting connection design from detailing across two organisations puts a boundary in the middle of the most iterative part of the work. It can be made to work, but only if changes travel as model changes rather than as markups on a PDF. Ask which it is.

5. "What happens to a superseded revision?"

The only acceptable answer is that it is retained and marked superseded. A supplier who overwrites is a supplier who cannot answer "what was the shop building to in March" once the job is finished and someone is arguing about it.

6. "Can we start with one sequence?"

A first scope small enough to check is worth more than any assurance. One erection sequence, one connection type, one lot — priced and delivered before the rest of the job depends on it. A supplier who resists a small first scope is telling you something.

What a bad answer sounds like

It is rarely a wrong answer. It is usually a fast, confident, general one: "we have a QA process", "everything gets checked", "we have done hundreds of projects like this". None of those can be checked, and none of them describes what happens on the specific Tuesday when a drawing is ambiguous and the schedule is tight.

A good answer is boring and specific, and it usually names an artefact — a numbered question list, an issue sheet, a revision register, a model check report. Artefacts are checkable. Assurances are not.

Run this list against us too

Five of the six have published answers. Questions two to five are at how we work — the numbered-question loop, the printed assumption, connection design running in the same model, and the rule that a superseded revision is retained rather than overwritten. The sixth is on for fabricators: a first scope small enough to check, priced and delivered before the rest of the job depends on it.

The first question is the one Seimora cannot answer yet. There are no completed projects under this name and so no job model to hand you, and work says that rather than borrowing somebody else's. Ask for the other five against a live enquiry. If any of those answers are worse than your current supplier's, that is useful information and it is better established now than on a job.

The same question runs the other way too — not what a shop should show you, but whether a standard decides who Seimora places work with — and how resources are selected publishes the same philosophy: a standard that can end in a no, not a checklist.