Retesting and Arbitration Testing: When to Ask, and How

Bottom line: separate repeat testing, retesting and arbitration testing

These three terms get used interchangeably in day-to-day conversation, but they differ completely in what triggers them, who initiates them, and how the outcome is relied upon. Mix them up and your request gets turned down, costing time for nothing.

• Repeat testing: the same laboratory, the same commission, one specific test run again. It normally follows an identifiable anomaly during testing, such as equipment status in doubt, a sample already damaged before the test, or a deviation from the plan. This sits inside the laboratory's own control of the validity of results, and in most cases the laboratory initiates it.
• Retesting: the sponsor challenges a result that has already been issued and asks for the retained sample, or a newly submitted sample, to be tested again against the original plan. It is a formal request. It needs a reason, it needs supporting material, and it has to be raised within the agreed period.
• Arbitration testing: two parties, typically a supplier and a customer, or a company and a regulator, disagree about the conclusion on the same batch of product, and a body both sides accept retests it against an agreed basis, with the result serving as the basis for settling the dispute. It presupposes identifiable parties to the dispute, an agreed basis, and usable samples.

The one-line rule: doubts about the conduct of the testing go down the retest route; disagreements between two parties go to arbitration; anomalies found inside the laboratory are handled by the laboratory under its control of the validity of results.

When a retest request stands up

The table below splits common situations into those that hold and those that do not. Check against it before you file and you will avoid a lot of pointless correspondence.

The row worth spelling out is "result is disappointing and you want another try." A retest is not another roll of the dice. If the product does not meet the requirement, the retest will say so too, and all you have bought is another invoice and another slot in the schedule. The right move at that point is to take the data back, find the failure cause, remediate, and test again under a new commission. Remediation is far cheaper and far faster than repeated retesting.

How to file a retest request properly

First, file within the deadline. Every laboratory has its own period for accepting challenges to results, usually written into the contract or the quality documents. Once it passes, raw records and retained samples may already have been dealt with under their retention schedules, and even a willing laboratory has nothing left to test. So check the report the moment you receive it, rather than going back to it only when a customer or a regulator raises a question.

Second, write down the technical grounds. Effective wording looks like "the test conditions for this item in the report differ from the plan description we approved" or "the report data differs from our witness records." Ineffective wording looks like "the result is unacceptable" or "our customer does not accept it." The first points at verifiable facts, the second points at feelings.

Third, submit the supporting material at the same time. Witness records, dispatch and packaging records, the signed plan, internal inspection data: as much as you have. The more specific the material, the faster the investigation.

Fourth, state what you are asking for. A review of the raw records? Retesting of the retained sample? New samples tested again against the original plan? Those three requests follow different paths and carry different cost allocations, and saying which one you want at the outset saves a round.

What the laboratory does with a retest request

Following the operating logic of ISO/IEC 17025, a laboratory receiving a complaint or an appeal runs a process independent of the personnel who performed the original testing: acknowledge, investigate, decide, respond, and take action where needed. A few points here are worth knowing.

• The investigation usually starts with the records, not with a repeat test. Raw records, equipment status records, environmental condition records and personnel authorization records are all pulled and checked. Many disputes reach a clear conclusion at this step with no retesting at all.
• The handling and the response are assigned to personnel who were not involved in the original testing. This is to avoid people reviewing their own work. If the response still comes from the same engineer, the sponsor is entitled to ask for the process to be followed.
• If the investigation confirms that the result was affected, the laboratory should manage the nonconforming work and take action on reports already issued. That may mean recalling the report, issuing a revised version, or issuing a supplementary statement. Conversely, if the investigation confirms that the original result stands, the laboratory issues its verification statement, and pressing further at that point achieves little.

Cost allocation is usually agreed like this: where the investigation confirms the cause lay with the laboratory, the retest carries no additional charge; where the original result is confirmed valid and the cause lay with the sponsor or with the product itself, the retest is charged as a new commission. Read this clause when you sign the contract, because arguing about it afterwards burns a lot of goodwill.

Arbitration testing: a higher bar than a retest

Arbitration testing is not "finding a more agreeable laboratory and testing again." Several conditions have to hold at once.

• Identifiable parties to the dispute. A supplier and a customer disagreeing on the same batch, or a company challenging the conclusion of a regulatory sampling inspection. One party feeling the result is wrong does not create an arbitration case.
• A body and a basis both sides accept. The basis for the inspection, meaning the standard applied, the acceptance wording and the sampling method, has to be confirmed in writing before work begins, or the two sides will still be talking past each other when it ends.
• Usable samples of clear origin. Normally drawn from retained samples of the original batch, or resampled in a way both sides confirm. If the origin of the sample cannot be explained, the conclusion binds nobody.
• An advance agreement on the outcome. Both sides need to agree beforehand that the arbitration result governs; otherwise the exercise just adds one more report to the pile.

In practice, the basis and the sampling are where arbitration gets stuck. What happens often is that the two sides are not actually arguing about data. They are applying different acceptance interpretations, one reading the product against one use scenario and the other against a different one. No amount of retesting resolves that; the basis has to be settled first. Settling the basis usually matters more, and costs less, than running the tests.

Three failure scenarios that will feel familiar

Scenario one: remembering the retest after the retention period. The report sits in a file for a long stretch, then a customer raises a question, and the sponsor goes back to the laboratory asking for a retest. By then the retained samples have been dealt with under the retention schedule; the raw records exist, but there is no sample to speak of. The only option left is new samples under a new commission, which means a new queue, a new invoice, and an extra justification for the consistency of the new samples with the original batch.

Scenario two: switching laboratories first, discussing the retest second. Unhappy with the result, the sponsor sends samples straight to another laboratory, gets a different conclusion, and comes back to confront the original one. The problem is that the two tests used different samples, possibly different plans and possibly different preconditioning, so the two reports are not comparable and the dispute becomes harder to close rather than easier. The correct order is to run the verification at the original laboratory first, then decide on arbitration once that conclusion is in.

Scenario three: using retests as a substitute for remediation. A result sits marginally outside the requirement, the team decides "maybe it passes next time," and several retests go by. Money and time are spent and nothing about the product has changed. A marginal result is precisely the signal that the design margin is thin, and at that point the return on spending the money on remediation and verification is far higher than on repeated retesting.

Killing disputes before the report is issued

Compared with retesting after the fact, it pays much better to deal with the things that cause disputes in advance:

• at plan approval, read the descriptions of test conditions and acceptance basis line by line and ask about anything unclear on the spot; do not assume "the laboratory knows what we mean";
• request witnessing for the critical tests, and keep your own copy of the witness record;
• when samples ship, keep records of sample condition, packaging and shipping method, photographs included;
• at draft report stage, check the model and specification, the sample description and the acceptance wording carefully, because changes cost essentially nothing at this stage and require a revision after signature;
• run screening tests internally so you know roughly what to expect before sending anything out. A clear gap between screening data and laboratory data often points early to a sample condition problem or a difference in interpretation.

For the pre-shipment preparation list, work against sample and documentation requirements; for the overall progression and where the time goes at each node, see our service process. Quick answers to common questions are in the FAQ.

A note on accreditation

In dispute handling, someone usually asks "you are an accredited laboratory, so how can there be a problem at all?" This needs saying plainly: an accreditation mark only demonstrates that the laboratory holds the corresponding technical competence within its accredited scope, and it is not a commitment as to market access outcomes. Accreditation addresses the conformity of competence and of the management system; whether the conclusion in any particular report is relied upon also depends on whether the test was within scope, whether the sample was representative, and whether the basis is accepted. Understanding that makes dispute conversations a great deal calmer.

Facing a disputed result? Do not ship new samples yet

If you are holding a report whose conclusion you doubt, send us the report and the related records first and we will help you work out which category the situation falls into, whether a verification route is available, and what material you need to prepare. In most cases the opening move should be a record review rather than a repeat test, and getting that step right saves a whole round of scheduling. SUNGO Lab (Shanghai Shage Medical Technology Co., Ltd.) is a medical device testing laboratory accredited by CNAS, CMA and IAS (USA), with laboratories in Shanghai and Hefei. Call +86 132 4819 8029 when you need assistance, describe the situation through technical support, or request a quote for a retest or a new commission proposal.