The Workbench · Craft
A concession can release product before testing ends
This blog has already covered how far a traceability procedure has to reach for an implantable device, and named, without stopping there, the neighboring clause a management review has to treat as its own standing agenda item: monitoring and measurement of product, listed in Clause 5.6.2 alongside monitoring and measurement of processes. Clause 8.2.6 is that clause on its own terms, and it does something a checklist mentality tends to miss: it doesn't just ask whether the product passed. It gates the release itself, and it names exactly one door around that gate — a documented concession, not a shortcut a supervisor waves through informally.
Release is a gate, not a formality
ISO 13485:2016 Clause 8.2.6 requires an organization to monitor and measure the characteristics of product to verify that requirements have been met, at appropriate stages of the realization process, in accordance with the documented procedures and the device's own acceptance criteria. The clause doesn't stop at running the test. It requires evidence of conformity to be recorded, the record to identify the person authorizing release, and — the sentence a checklist-style file tends to compress into a single checkbox — release and delivery of product not to proceed until the planned arrangements have been satisfactorily completed. A record that shows a passing test result but no separate field for who authorized the release, and on what evidence, has recorded a measurement, not the release decision the clause actually requires.
Implantable devices add a name to the record
For an implantable device, Clause 8.2.6 goes further: the organization has to record the identity of personnel performing any inspection or test. That's a narrower, stricter duty than the general release record, and it sits next to the traceability reach this blog has already covered for implants — both provisions single out implantable devices for a level of individual accountability the standard doesn't ask of every device class. A release record for an implant that names the test but not the tester has satisfied the general population of Clause 8.2.6 while missing the specific sentence the standard wrote for exactly this device category.
A concession is the other door
The clause's own release sentence carries a conditional clause worth reading closely: product isn't released before the planned arrangements are satisfactorily completed, unless otherwise approved by a relevant authority and, as applicable, by the customer. That's not a loophole the standard leaves open by accident — it's a deliberate second path, narrower than the first, for the case where a device needs to move before every planned check has actually returned a result. A familiar version of exactly this concession shows up around sterilization: a lot released to distribution while a biological-indicator or sterility result is still pending, on the strength of a validated cycle's own process parameters rather than the completed test that would otherwise close the gate. This blog has already covered why a validated process substitutes for inspection in the first place — a concession built on that validation is where the two ideas meet directly.
A concession is not a waiver of the criteria
What the concession changes is the sequence, not the standard. The acceptance criteria the device has to meet don't move, and the underlying evidence still has to arrive and still has to pass — the concession only authorizes the release to happen before that evidence is in hand, on the strength of a documented approval from a relevant authority and, where the arrangement calls for it, the customer. A record that shows “released, results pending” with no approval on file has recorded a shortcut nobody with the standing to authorize it actually signed off on, whatever the outcome of the pending test turns out to be. The approval is what makes the sequence lawful; the eventual passing result is a separate fact that doesn't retroactively supply it.
Where this meets the file
A release record built around this clause needs three fields most trackers collapse into one: the acceptance-criteria evidence itself, tied to the specific stage it closes; the identity of the person authorizing release, plus the tester's own identity where the device is an implant; and a distinct concession field, populated only when release happens ahead of completed evidence, naming who approved it and under what arrangement. A device-release worksheet built around that structure, alongside the traceability and process-validation logic this blog has already covered, is previewed in the launch catalog. If your quality system runs concessionary release differently, the shelf takes that correction directly.
The Regulatory Toolkit launches soon — a free shelf of source-mapped templates, checklists and browser-only tools for regulatory teams. Get one email when it opens, or contribute a template.