Talk · Compliance · Test reports, labels & documents
After a failed test, when is a retest enough and when is a redesign needed?
A retest is justified only if you can show that the failure came from a test or sample error and not from the design.

One of our samples failed a lab test. Can we just retest, or do we need to redesign?
Short answer: A retest is justified only if you can show that the failure came from a test or sample error and not from the design. If the cause is in the design, parts or process, retesting the same product wastes time and may fail again. Find the root cause first, fix it, document the change and then retest the fixed version. Margin matters: a result near the limit is a warning.
Start with root cause
Do not retest first. Ask why it failed. Look at the data: by how much did it miss the limit, at which point, and was the result repeatable?
When a retest may be reasonable
- Test setup error, such as a wrong cable, load or connection, and the lab agrees.
- A faulty sample that was damaged in handling or was not built to specification.
- Measurement problem, such as a calibration issue.
- A marginal result with a clear explanation, confirmed with a second sample, and agreement from the lab or certification body that retesting is permitted.
When a redesign or change is needed
- A consistent failure across samples.
- A large miss from the limit.
- A design cause, such as a poor layout, a weak component or a missing protection.
- A process cause, such as poor assembly or soldering.
- A near-limit pass that leaves no margin for production variation.
Steps
- Collect data: the report, the setup, the sample and the measurements.
- Reproduce the failure in your own lab or the factory’s.
- Find the root cause, with the designer.
- Decide on the fix: a part change, layout change, firmware change, shielding or a process fix.
- Assess the impact on other tests and approvals.
- Update the drawings, parts list and revision number.
- Build fixed samples and test internally with margin.
- Inform the lab or certification body and agree on the scope of the retest.
- Retest and record the result.
- Update your records and apply the change to production.
What not to do
Do not send the same unit again hoping for a different result, and do not change the product without recording it. Do not hide a failed test from the certification body.
Keep margin in mind
A product that passes by a tiny margin may fail on the next batch. Aim for a safe gap.
Quick summary
Find the cause. Retest only if the cause is not the design, otherwise fix, document and retest.
Where this answer says to confirm with Wecent, the detail is not published on the product page and has to come from the datasheet or the factory. General facts come from the regulators, standards bodies or device makers named in the text.
Three more perspectives
Products discussed in this thread
Related articles & topics
Have a sourcing question about your own programme?
Ask our team More Test reports, labels & documents questions