Testing Laboratory and Materials Testing Insurance

This is the most important distinction in testing liability and it needs to be visible on every certificate.

A laboratory tests what arrives. If a contractor's site engineer made the cubes, cured them in a puddle by the gate, took a soil sample from the wrong depth, or sent a water sample in an unsuitable container, the result is a correct measurement of a sample that does not represent the structure. The contractor, the designer and eventually a lawyer will treat the certificate as a statement about the concrete, not about the cube.

So the certificate has to say what it is. Who took the sample, when, where, under what method, and the condition it arrived in. Record non-conformities rather than absorbing them: a cube that arrived damaged, a sample outside its holding time, a container that was wrong, a chain of custody that is incomplete. Reporting a result with a stated non-conformity is professional; reporting it silently is how a laboratory ends up answering for somebody else's sampling. Where you do the sampling yourself, that is a different and larger exposure, and it should be insured as such.

The reason testing carries a loss out of all proportion to its fee is that results are acted on at once and never revisited.

A concrete cube result at seven or twenty eight days tells a contractor whether to strike formwork or load a slab. A soil classification sizes a foundation. A weld inspection accepts a connection. An air test completes a handover. A water result releases a supply. In each case the consequence happens within days and nobody tests again, so an overstated result is discovered later, through a failure, a defect or an audit, with the structure already in place and following trades finished.

Which puts the emphasis on the things that prevent a wrong number leaving the building: calibration within date and traceable, method followed rather than adapted, a second check on results that are unusual or that sit close to a specification limit, and a culture where a technician can flag a result that looks wrong without it being a problem. The last one matters most, because the results that cause claims are usually the ones somebody felt uneasy about.

There is one failure mode in a laboratory that produces not one claim but every claim since the last good check.

A compression machine reading high, a balance out of adjustment, a thermometer drifting in a curing tank, an oven not holding temperature, or a reference standard past its date does not produce one wrong certificate. It produces every certificate issued since the equipment was last verified, which may be hundreds, across many clients and many projects. When it is discovered, the laboratory has to decide what to tell whom, and the exposure is a notification exercise rather than a single dispute.

So calibration and verification records are the business rather than an administrative burden: intervals adhered to, intermediate checks between calibrations so drift is caught early, records traceable per instrument, and a documented procedure for what happens when equipment is found out of tolerance, including how far back results are reviewed and who is informed. An insurer will want to know that procedure exists, because a laboratory that can bound the problem quickly is a very different risk from one that cannot.

Laboratories get drawn into saying what results mean, and that crosses a line between two different professional standards.

Reporting that a cube achieved a stated strength, that a soil has a given classification, or that a weld shows an indication of a certain size is testing. Saying that the slab is therefore safe to load, that the foundation design is adequate, that the weld is acceptable for the application, or that a site is suitable for a use is engineering interpretation. Clients ask for the second constantly, frequently in an email, because it is what they actually need.

So decide where the line is and hold it in writing. State on reports that results relate to the samples tested and that interpretation for design purposes is a matter for the appointed engineer. Where you do offer interpretation, price it, scope it and insure it as advice rather than giving it away in a reply. And where a client's question implies a decision you are not qualified or appointed to make, say so rather than being helpful, because a helpful sentence in an email is exactly the document that gets produced later.

The limit is the question, because the loss is the structure. We are an FCA regulated broker and will not tell you we are the best choice. These are the questions that decide it.