What a failed test letter should tell a customer about a sample
Summary
A failed sample letter takes five steps: result against limit, test conditions, what the failure means, next steps, and removing internal notes. They apply when a supplier or laboratory tells a customer that a chemical or lubricant sample missed a numerical specification. Near the limit, the letter also names the decision rule that turned the value into a verdict.
ISO/IEC 17025:2017, Edition 3, was last reviewed and confirmed in 2023 and is the active edition of the general requirements for testing and calibration laboratories (ISO, 2023).
Under simple acceptance, a result exactly on the tolerance limit may have up to a 50% probability of being outside it, assuming a symmetric normal distribution (ILAC-G8:09/2019).
ASTM D3244-24, the active edition last updated 19 December 2024, resolves supplier and receiver disputes by comparing an assigned test value from both laboratories with an acceptance limit (ASTM D3244-24, 2024).
ISO 4259-2:2026, valid from 9 June 2026, determines specification conformance status when supplier and receiver results conflict for petroleum and related products (ISO 4259-2:2026).
This guide covers letters about a sample that failed a numerical specification for a chemical or lubricant. It does not cover regulatory non-compliance notices, recalls or contract claims, which follow their own legal rules. Nor does it replace the laboratory's own test report.
How do you state a failed result against the specification limit in a customer letter?
State the failed result in one sentence that holds the measured value, its unit, the specification limit, the test method with its edition, and the specification the limit comes from. ISO/IEC 17025:2017 clause 7.8.6.2 requires a laboratory's statement of conformity to identify to which results the statement applies, which specifications are met or not met, and the decision rule applied, unless it is inherent in the specification (ILAC-G8:09/2019). A customer letter is not a laboratory report, but the same three points answer the customer's first questions.
An illustrative sentence: "Sample 24-118 measured 520 mg/kg water by the method named in your specification, against a maximum of 500 mg/kg, so it does not meet clause 4 of that specification."
Round the result the way the specification states it. ASTM D3244-24 recommends, in the absence of an agreement, rounding by the rounding-off method of ASTM E29 to the number of significant digits in the governing specification (ASTM D3244-24, 2024). Put the attached laboratory report number next to the sentence, so the customer can check the value against the source document.
Leave out words that soften or harden the number, such as "slightly over" or "grossly out". The value and the limit already say how far apart they are.
Which test conditions belong in a failed sample letter?
Include the test conditions the customer needs to match the result to its own records and to judge whether the test was right for the material. Five items do that work: the sample identity in the customer's own terms (their order, lot or container reference), the date the sample was taken and the date it was received, who took the sample, the laboratory and the test method with edition, and any deviation from the method.
Who took the sample matters because a result describes the material that reached the laboratory. Perry Johnson Laboratory Accreditation's summary of ISO/IEC 17025:2017 notes that when the sample is provided by the customer, the report shall state that the results apply to the sample "as received" (Perry Johnson Laboratory Accreditation, 2019). If the sample arrived in an unusual condition, name it: a container that was not sealed, for example, or a sample older than the method allows.
If the laboratory is accredited for that method, say so; if it is not, do not imply it. Keep units and method names exactly as written on the laboratory report. A customer comparing documents will treat any difference as a second, unexplained result.
How do you explain what a failed sample result does and does not mean?
A failed result means that this sample, as received and tested by this method, did not meet this limit. It does not by itself show why, whether other lots fail, or whether the product will fail in use. Write both halves into the letter, because a customer who reads only "failed" will assume the worst.
What decides the wording is how close the result is to the limit compared with the measurement uncertainty. ILAC-G8:09/2019 notes that under simple acceptance, where the acceptance limit equals the tolerance limit, a result exactly on the limit may have up to a 50% probability of being outside it. ISO/IEC 17025:2017 clause 7.8.3.1 c) has test reports include measurement uncertainty when it affects conformity to a specification limit (ILAC-G8:09/2019).
The JCGM guide on conformity assessment states that choosing tolerance and acceptance limits is a business or policy decision (JCGM 106:2012), so the letter should say which rule was used rather than argue the rule.
The table below sets what the letter can claim. It uses the ILAC-G8 non-binary statements, which add a guard band w, a margin above and below the limit, to plain pass and fail. These rows apply only when a non-binary decision rule was agreed with the customer before testing, as ISO/IEC 17025:2017 clause 7.1.3 requires (ILAC-G8:09/2019); under simple acceptance, a result above the limit is a fail.
Where the result sits | ILAC-G8 statement | What the letter can say |
|---|---|---|
Above the limit plus guard band | Fail (ILAC-G8:09/2019) | The sample does not meet the limit |
Between the limit and limit plus guard band | Conditional fail (ILAC-G8:09/2019) | Above the limit, within measurement uncertainty; propose a retest |
Between limit minus guard band and the limit | Conditional pass (ILAC-G8:09/2019) | Not a failure; note it is close to the limit |
The conditional fail row needs the most care. There the letter reports the result as above the limit and proposes a retest instead of a final verdict.
What next steps should a failed sample letter propose?
Propose one to three next steps, each with an owner and a date, so the customer knows what happens and who acts. Typical steps are a retest of the retained sample, a referee test at a second laboratory, a hold on the affected material, and an investigation of cause with a date for the first finding.
For petroleum products and lubricants, ASTM D3244-24 describes how two parties compare and combine independently obtained results into an assigned test value and compare it with an acceptance limit. The practice states that its application "requires the AL be determined prior to actual commencement of testing", and recommends that it be conducted under the guidance of a qualified statistician (ASTM D3244-24, 2024). ISO 4259-2:2026 covers the same supplier and receiver conflict for petroleum and related products, which are normally homogeneous (ISO 4259-2:2026). So if a referee test is likely, propose that both parties agree the method, the laboratory and the acceptance limit before any sample is retested.
Do not promise a cause or a corrective action in the first letter. Promise the date when the first finding will be shared. Then write who decides what happens to the material on hold, because the customer cannot plan production until that is known.
What internal information should stay out of a failed sample letter?
Keep out the working history of the case, internal review comments, confidence scores or internal ratings, internal record sources such as chat threads or CRM entries, and data about other customers. These items describe how the supplier reached its view, not the result, and a customer can read them as admissions or as doubt.
Take a comment such as "lab said this method is unreliable" in a forwarded thread. It undermines the result the letter reports, without the evidence that would let the customer judge it. Replace each internal source with a customer-held document: the laboratory report, the certificate of analysis or the specification.
Uncertainty belongs in the letter only as the laboratory states it, with the method and unit, not as an internal confidence figure. Cost and margin rarely belong in a failed sample letter. If a credit or replacement is offered, write it as a commercial decision, separate from the technical result.
The guide on what to leave out of a technical report for a customer sets out the full rule set for customer documents, item by item. Before sending, read the final letter once as the customer would, and ask of each sentence whether it states a fact the customer can check.
How do you check a failed sample letter before it goes to the customer?
Check the letter line by line against the attached laboratory report, with a second person who did not write it. The table below decides whether the letter can be sent or what to fix first.
Check | Pass | Fail and what to do |
|---|---|---|
Sample identity and dates | Match the customer's order, lot and sampling records | Correct from the laboratory report |
Result and limit | Value, unit, limit, method and edition in one sentence (ILAC-G8:09/2019) | Rewrite the sentence from the report |
Decision rule near the limit | Rule or measurement uncertainty stated when it decides the verdict (ILAC-G8:09/2019) | State the agreed rule; if none was agreed, report the result with its uncertainty and propose a retest, without relabeling the verdict |
What the failure does not show | Written in plain words | Add one sentence on scope |
Next step | Owner and date for each step | Name both before sending |
Internal content | No history, scores or internal sources | Remove and cite customer-held documents |
The row most likely to matter later is the decision rule near the limit. A result within measurement uncertainty of the limit invites a retest, and the letter should already have offered one.
When Lubinpla's document writing prepares a document for a customer reader, it leaves out work history, confidence and scores, and internal record sources. Send the letter as a PDF with the laboratory report attached, not as text pasted into an email thread, so the reply chain does not carry earlier internal messages.
If the customer questions the method or the rule, route the question to the issuing laboratory and confirm its reply in writing. Keep the sent version and the laboratory report together in the case file.
Frequently asked questions
Should a failed sample letter apologize?
A failed sample letter can acknowledge the impact on the customer without apologizing for a cause that has not been found. Thank the customer for the sample, state the result, and promise the date of the first finding. An apology for a confirmed cause belongs in the follow-up letter.
Who should sign a failed sample letter?
The person who owns the customer relationship usually signs, with the technical contact named for questions about the method. If the laboratory issued a statement of conformity, the letter should not change its wording; quote it and attach the report.
Can the letter say the product is still fit for use?
The letter can call a product fit for use only if someone with the evidence and the authority has decided that, and the letter names the basis. Otherwise, state the result, propose the next step, and leave the use decision to the agreed process.