When a COA Doesn’t Add Up: How to Demand Real Answers
A certificate earns trust only when it traces to the lot in your hand. Here is how UK buyers audit a certificate of analysis, write a clarification request, and judge whether a re-test is the honest answer.
A certificate of analysis arrives, and one field is blank. A method line reads “in-house” with no instrument, column, or parameters. A purity figure will not reproduce on your own check. This is when a document stops being paperwork and becomes a question you must ask in writing.
Trust in a certificate of analysis does not come from a clean PDF. It comes from traceability to the specific lot in your hand. When a field is missing, generic, or inherited from an earlier batch, the right response is a documented clarification request and, where warranted, a re-test of the retained sample, never a re-dated copy.
This guide walks through the work in order. You will learn what a certificate of analysis actually asserts, the records to gather before you query anything, how to audit the certificate field by field, and how to classify a defect before you write. It then covers drafting a clarification request that cannot be answered with a template, judging whether a reply is a legitimate reissue or a re-dated copy, and recognising when a retained-sample re-test is the only honest option. It closes with the questions buyers ask most and the red flags that stall a supplier.
Why a Certificate Only Earns Trust When It Traces to Your Lot
A certificate of analysis arrives and something does not sit right. A result field is blank. The method line reads only “HPLC”. A purity figure will not reproduce when you check it against the material in front of you. That moment of doubt is where the document stops being evidence and becomes a question.
Trust is a property of that link, not of the document itself.
Traceability is the working test. Four elements must appear and must match what you received: the lot or batch number, the test date, the method reference, and a result stated against a specification. All four should reconcile with the vial, the label and the invoice.
The failure this article targets is rarely forgery. It is inheritance: a sheet carried over from an earlier batch, or a template with fields left empty because nobody checked them against the lot that shipped.
If you want a field-by-field walkthrough of what a peptide certificate should contain, our guide to reading a peptide certificate of analysis covers the same four checks. The rest of this article shows how to challenge a certificate that fails them.
What a Certificate of Analysis Is Actually Asserting
A certificate of analysis is not a badge or a marketing asset. Treated as a test report, it should assert five things at once: a result, produced by a method, by a laboratory, on a sample, at a date, judged against a specification. Remove any one and what remains is not evidence.
That structure is why testing laboratories are assessed against ISO/IEC 17025, and why ANAB has treated the evaluation of metrological traceability on certificates as a subject in its own right. Buyers are expected to interrogate the chain behind a number, not simply read it. This article cites that framing only, not clause text. Where your process needs specific requirements, read them from the standard itself, not from any supplier summary, including this one.
The same logic appears in regulation. NIBSC guidance for licensed biological medicines treats the individual batch, not the product family, as the unit requiring certification before UK market entry, with notification at least seven working days before a batch is placed on the market and a seven working day MHRA objection window. That is procedure for licensed biological medicines, used here strictly as an analogy for why batch-level documents matter. It does not bind research-grade peptides, and nothing here should be read as saying it does.
Strip the regulation away and the logic still holds: a document that describes a product type rather than one lot cannot tell you anything about the material in your hand. Any Certificate of Analysis you accept should therefore be readable as a statement about one lot, with a defined method and a defined specification attached to it.
Prerequisites: The Records You Need Before You Query Anything
Those records rarely sit in one place. Assemble these four before you query anything, then fix two definitions.
The lot or batch number in every version it appears. Copy it exactly as printed on the vial or label, then as it appears on the invoice and on the certificate. Three renderings of one identifier is itself a finding worth writing down, and what a batch record can and cannot tell you explains why.
Order date and date received. These let you test whether the test date on the certificate plausibly precedes dispatch. A test date after delivery is not impossible, but it demands an explanation before the document can be relied on.
The certificate as originally supplied to you. Save it unaltered, with the filename and the date you downloaded it. If the original is lost, a later replacement has no audit value: you cannot show what changed, only what now appears.
The specification the material was sold against. A purity figure, an identity requirement or a stated assay range. A result is meaningless without the target it is compared to.
Two definitions are worth fixing before you start. A batch-specific certificate of analysis is tied to one lot by identifier; a generic or representative certificate describes the product type. A specification is the limit the result is judged against, not the result itself.
With those in hand, you can audit the certificate field by field rather than argue about impressions.
Step 1: Audit the Certificate of Analysis Field by Field
With those records in front of you, audit the certificate in a fixed order. Five checks, each a question the document must answer before it counts as evidence.
- Lot number. It must name the lot in your hand. A catalogue code, a product family name, or wording such as “all batches” or “representative sample” fails this test outright.
- Dates. Test date and report or issue date are separate fields. A report issued long after the test, or a test date that postdates your dispatch date, needs an explanation before the document can be relied on.
- Method. “HPLC” or “mass spectrometry” alone does not identify a method. Look for system, column and detection detail sufficient for a competent laboratory to repeat the procedure, plus the acceptance criteria applied to this lot.
- Result against specification. Every measured value must be paired with the limit it is judged against. “98.7%” alone is a number in space; “98.7% against a specification of not less than 98.0%” is a pass you can act on. That pairing principle runs through specification guidance for drug substances, including the ICH Q6A guideline, which governs drug substances, not research-grade intermediates, but the underlying logic holds for any batch certificate.
- Identity, laboratory and reference. Check identity evidence alongside the purity figure, who performed the testing and under what quality system, and whether the report carries a unique reference you could quote back in a written query. The same audit covers related questions the certificate should already settle.
Log each failed check in writing now. Deciding what kind of defect you have, and what to ask for, is the next step.
Step 2: Classify the Defect Before You Write to Anyone
Four categories cover almost every defective certificate, and the label you apply decides what you ask for next.
Missing. A required field is blank or absent: no lot number, no test date, no specification line, or a result field with nothing in it.
Generic. The document is written at product level, describing the material type rather than your lot. Nothing on the page is specific to your sample, so nothing on it can be read as evidence about your vial.
Inherited. The values look carried over from an earlier batch. Telltales: an old lot number left in the header, an unchanged test date, or results identical to a certificate you already hold from a previous order.
Contradictory or non-reproducing. The stated result does not sit against the stated specification, the arithmetic of a purity breakdown does not close, or your own check, or an independent laboratory’s, returns a materially different figure. The FDA’s non-binding out-of-specification guidance for pharmaceutical production calls such a result out-of-specification, and that moves the issue from documentation to sample.
Classification decides your next move. A missing field is usually answered by a reissue; an inherited certificate usually is not, because it needs data for your lot that may never have been generated. A non-reproducing figure triggers a re-test conversation about the retained sample.
Write nothing until you have chosen the label. The wrong request invites the wrong reply.
Step 3: Write the Clarification Request That Cannot Be Answered With a Template
With the defect classified, the email writes itself. Open with identifiers, not frustration: product name, catalogue reference, lot number, order number, date received, and the certificate reference you are querying, quoted exactly as it appears. Then state the ask in one line so it reaches the right desk without reinterpretation. Related questions on reading these documents are covered in how to read a peptide certificate of analysis.
Ask for the lot-linked document
Write: “Please provide the certificate of analysis issued for lot [X], not a representative certificate for this product.” If what you hold is generic, say so, and ask whether a batch-specific certificate for your lot exists at all.
Name the method parameters
Ask for the chromatographic system or column, mobile phase or gradient, detection mode or wavelength, injection volume, and the acceptance criteria applied to this lot: “Please confirm the method parameters used to generate the result reported for lot [X].”
Request the underlying data
“Please provide the chromatogram, the mass spectrum or the supporting test report for lot [X], or confirm in writing that these are not available.” Written confirmation of unavailability is itself a useful record; a supplier holding the data will produce it.
Close with the questions templates cannot answer
“Confirm whether this lot was tested against a specification, and state the specification applied.” Then: “Confirm whether a retained sample of lot [X] is held, at what quantity, under what storage conditions, and for how long.” Set a response window and state what you will do if it lapses, so the email becomes a record rather than a favour.
Step 4: Judge the Reply: Legitimate Reissue or Re-dated Copy
Once the reply lands, judging it is mechanical. Compare the document you were sent against the one you queried, field by field.
Good document-control practice requires a legitimate reissue to carry a new issue date alongside the original, a stated reason for reissuing, and a version or revision marker. That keeps the document’s history visible to anyone reading it later. Anything less hides the change rather than recording it.
The values must be your values. If a reissue repeats an identical test date, identical results and the same report number but arrives with a different lot number, you are holding a re-labelled sheet, not new evidence.
Keep one distinction firmly in view: a reissue is not a re-test. Correcting a typographical error or filling a field left blank is a documentation fix. A documentation fix does not generate new analytical data, only a re-test of the sample can do that.
Method answers are adequate only when another laboratory could repeat the work from them. If the reply restates “HPLC analysis confirmed purity” in different words, the question has been deflected rather than answered.
So ask directly: “State what was amended, who authorised the amendment, and confirm that the sample tested is from lot [X].” Where that cannot be answered, the document has not been validated against your lot, whatever it says at the top. Our answers and resources set out what a defensible test report contains if you want the field-level detail behind these questions.
Step 5: When a Retained-Sample Re-test Is the Only Honest Option
When a result will not reproduce, paperwork stops being able to settle the question. A field that is merely missing can be answered with a document; a material discrepancy between the certificate and your own check, or an independent laboratory’s check, cannot. That is the trigger for a retained-sample re-test.
The retained sample is the mechanism. A portion of your lot, held by the supplier or by you, can be tested independently without opening or consuming the material in use. Knowing whether one exists matters before you need it. The cGMP reserve-sample framework addresses what a defensible retained sample looks like: a lot-representative portion, stored under defined conditions for a defined period. That framework does not bind research-grade peptides, but it shows the standard to aim for.
Put the practical criteria in writing:
- Whether a retained sample of lot [X] is held.
- The quantity held.
- The storage conditions.
- How long after dispatch it is kept.
Storage condition matters, because a sample held badly is not a sample you can rely on.
Who pays for the re-test, who performs it, and how the result is reported are legitimate questions. Ask before agreeing to anything, and keep the answers in the batch record, not an email thread.
Plan for three outcomes:
- The re-test confirms the certificate. Record the resolution.
- The re-test contradicts it. This opens a replacement or credit discussion; see batch quality concerns and recalls before agreeing terms.
- The supplier refuses to re-test. Refusal is also a finding: it shows where that supplier’s quality position actually sits.
Step 6: Close the Loop and File the Batch Record
Once the re-test decision is settled, the file is what keeps it settled. Close the loop by assembling the set, not the sheet.
File the set, not the single document
Attach the certificate of analysis as it was originally supplied, your written clarification request, the supplier’s reply, and any re-test report or written refusal to re-test. Store the outcome against the lot number, so a future query about the same lot starts from evidence rather than from memory or a colleague’s recollection.
Record response time and reply quality
Log how long the supplier took to respond and whether the reply actually answered the question you asked. It is the most practical measure available of whether the quality system behind a certificate is doing real work at batch level.
Keep your own copy, always
Save the version you were supplied with even where the certificate is published online. A page that changes silently leaves you holding a record you cannot later produce alongside the material.
Audit every supplier the same way
Apply the field-by-field check regardless of who issued the document. Apply these four checks to every certificate, regardless of supplier.
Tips: Phrases That Stall a Supplier and Red Flags in Replies
Filing the batch record is only as useful as the questions that produced it. The wording you choose decides whether a supplier can answer with a template.
Phrases that stall. Drop “please send more information”, “can you confirm this is genuine” and “any details would help”. Each invites a paragraph that says nothing. Replace them with closed questions tied to named fields: “State the specification applied to lot [X]. Confirm the method parameters used to generate the purity result. Confirm whether a retained sample is held.” A non-answer is only visible when the question was specific enough to be answered.
Never accept “representative”. A representative certificate describes a product, not your lot. Ask in writing for the lot-linked certificate of analysis, and if the reply is a refusal, keep it on file with the date and the exact wording. A refusal is a record.
Red flags in replies:
- No specification. Results without a stated limit cannot pass or fail, so every number looks acceptable.
- Reused documentation. An identical test date across multiple lots, or a report reference that recurs across separate shipments, suggests one sheet is doing duty for several.
- Marketing in place of method. If the reply describes the testing process in promotional terms rather than giving parameters, the question has been deflected, not answered.
Silence is also a red flag, and it is a result. Where a supplier will not answer a batch-level question in writing, the sensible next step is to source the material from someone who will.
Certificate of Analysis Questions Buyers Ask Most
Once you have filtered out the stall tactics and red flags, five questions recur. Here are the answers that separate a lot-linked certificate from a template.
Which fields must be batch-specific on a certificate of analysis? See Step 1 for the full field-by-field audit; in brief: the lot number, the test date, the method reference, and the result stated against a specification must all be lot-specific. Other content can legitimately sit at product level.
How do I request method parameters and raw data? Ask by name for the chromatographic system or column, the mobile phase or gradient, the detection mode, the injection volume and the acceptance criteria. Then ask whether the chromatogram, the mass spectrum or the supporting report is available for your lot. Named items make a partial answer obvious.
How do I tell a generic certificate from one tied to my lot? Search the document for your lot number and your test date. If either is absent, or if the same sheet is offered to you for several lots, it is not lot-linked.
Does a reissue mean the material was retested? No. A reissue can correct a field, a typographical error or a missing line. Only a re-test produces new data, and only a re-test of the retained sample is the honest request when a result will not reproduce.
What should I do if the supplier refuses? Record the refusal with the date and the exact question asked. Then decide whether to retain the lot. A refusal to answer a batch-level question is information about the supplier, not a dead end.
With these answers, the certificate becomes a document you can test against your lot, not just read.
The Certificate Is Only as Good as Its Link to Your Lot
Those answers only matter when applied to the document in front of you. Keep one rule in view: a certificate of analysis earns trust only when it is traceable to the specific lot in your hand, not to a product name or a catalogue code.
Apply these four checks to every certificate, regardless of supplier.
A re-dated copy of the same sheet is never evidence about your lot. If the four fields cannot be tied to the material on your bench, the certificate has not earned its place in your batch record.
Conclusion
Your next step is simple: run the four-field audit on every certificate, from every supplier, before you file it. If the fields cannot be traced to your lot, request clarification or reissue, make traceability the standard, and your batch records will stand up when it matters.