Product testing

Certification runs on laboratory evidence, and whether a report is usable is decided by two things that have nothing to do with how good the testing was: whether the laboratory is accredited for that specific standard, and whether the destination recognises that accreditation. Accreditation is granted per standard and per laboratory, never generally, so a laboratory accredited for one product standard is not automatically accepted for the next one. We specify the test plan, place the work with laboratories the destination accepts, and where a product is going to several markets we specify one campaign that supports all of them rather than one per market.

A report is only worth the accreditation behind it

The most expensive avoidable cost in certification is re-testing, and it is almost always caused by the same thing: work placed with a laboratory that was competent but not accredited for the standard in question, or accredited by a body the destination does not recognise. The report comes back clean, the certification body refuses it, and the samples ship again.

The recognition question is separate from the accreditation question and has to be asked second. ISO/IEC 17025 accreditation by a recognised national body is the baseline, but several schemes maintain their own lists on top of it — the Eurasian Economic Union requires a laboratory entered in its unified register, and a report from an ISO/IEC 17025 laboratory outside that register is not usable for an EAC document however sound it is.

Where a scheme operates a recognition arrangement, using it is the cheapest route available. The IECEE CB scheme is the clearest example: a CB test certificate and report commissioned once are accepted, subject to national differences, by the member schemes rather than being repeated in each. That only works for a manufacturer who commissioned the testing as a CB project in the first place, which is a decision taken before the first booking.

Test once for several markets, or pay per market

Markets that reference the same underlying standards can frequently be served by one campaign. The saving is real and it is entirely a function of planning: the test plan has to name every intended destination, so the clauses each of them cares about are covered in one pass and the report is written in a form they all accept.

Where that planning has not happened, the second market is not half price. The reports exist, they were scoped for one destination, and what is missing is precisely the delta the second destination requires — which usually means new samples, a new booking and a new queue.

The queue is the other thing worth planning around. For most product families the elapsed time in this stage is not the test, it is the wait for a slot, and slots are booked weeks out. Sample availability is the reciprocal problem: certification samples have to be production-representative, and a hand-built pre-production unit that differs from what will ship is a finding rather than a shortcut.

What is priced, and what surprises people

Testing is priced per standard and per sample rather than per product, which is why a small range can cost more than a single complex device. Textiles are the case that catches buyers out most often: testing is priced per material and per colourway rather than per garment, so a six-colour range is six times the work on the parts of the standard that depend on dye.

Destructive testing is the other line item people underestimate. Several safety standards consume the sample, and a product tested to a full suite may need several units. Where samples are expensive, the test plan can often be sequenced so the destructive work happens last, after the cheaper tests have had a chance to fail.

What decides whether a report will be accepted

The left-hand column is what a first estimate assumes and the right-hand column is what the invoice reflects. Every row in it is knowable before the first booking.