Thirty Questions Before You Qualify a Bile Acid Supplier

Supplier qualification usually begins with a questionnaire and ends with a price. The questionnaire is long, the answers are reassuring, and nothing in the process establishes whether a number on a certificate can be traced to a measurement.

This list is shorter and harder. Thirty questions, six blocks. Each one is answerable, and each answer either holds together with the others or does not. We wrote it as the list we would want a customer to ask us, in this order.

How to use it

Send the first two blocks before the commercial conversation. They can be answered by email and they determine whether the rest is worth the time. Keep blocks three to five for the site visit or the document review, because those answers are checked against records rather than heard. Block six belongs in the contract discussion.

One rule makes the list work: ask for a document, not a statement. A supplier who answers with records has the records. A supplier who answers with assurances is describing intent, and intent is what an audit tests.

Block 1 — Identity and specification (Q1–Q5)

1. Which specification governs this material, and what version is it? A framework name is not a document. The answer should identify a revision, or state that a company specification is applied and how it relates to the compendial text.

2. Where did each limit come from? Compendial, derived from the site's own batch distribution, set by a customer, or inherited. The fourth answer is common and should be admitted.

3. Which tests are additional to the compendial set, and why? Additional tests are a signal of process knowledge. A specification that exactly mirrors a monograph has told you nothing about the site.

4. Which parameters are not routinely tested? The most informative question in the block. Elemental impurities, nitrosamine risk, and microbial counts beyond the compendial set are common exclusions, and each one takes time to add.

5. Is the material described by a monograph, or aligned with one? The difference decides what the customer's regulatory file can claim, and it is frequently blurred in correspondence. The six statements that close this gap are worth having in writing before the first order.

Block 2 — Methods and data (Q6–Q10)

6. What method produced the assay figure? Titration and chromatographic comparison differ by more than half a percent on the same lot, so the number is not portable without the method.

7. What is the intermediate precision for that method, across days and analysts? A standard deviation, not a specification. Without it, no difference between two certificates can be judged.

8. What is the expanded uncertainty, and which term dominates it? Usually the reference standard's assigned purity. A supplier who knows this has built a budget; one who quotes a figure without a dominant term has inherited it.

9. Which test results come from a third-party laboratory, and is it accredited? Not a criticism of in-house testing. A question about which results are external, because that determines what the customer can rely on when the two laboratories disagree.

10. Can you provide the chromatograms, not only the integration result? The trace shows whether the baseline was placed consistently and whether small peaks were seen. This is the level at which integration decisions become visible.

Block 3 — Traceability and records (Q11–Q15)

11. Where in the process is the batch number assigned? At synthesis or at packing. The answer determines whether the number indexes a production history or a warehouse movement.

12. Does repacking create a sub-lot identifier? If portions can be dispositioned or stored differently, one number cannot describe them both. The traceability boundary sits exactly here.

13. Can you produce the batch record for a named batch during an audit? The only question in this block that tests the system rather than the label.

14. How long are records retained, and in what form? Long enough to cover the material's shelf life plus the customer's filing period. Paper archives that cannot be retrieved are not retention.

15. How is material from a previous campaign segregated from current stock? Physical segregation and record segregation are different controls, and the second is the one that fails.

Block 4 — Change and stability (Q16–Q20)

16. What counts as a change requiring notification, and how many days' notice? The customer's regulatory clock runs slower than the supplier's production clock, and the notice period has to respect the first.

17. Has the process changed in the last three years? What changed? A site with no changes in three years is either stable or not recording them.

18. Which stability condition was used, and does it match our market? A study at 25 °C and 60% relative humidity answers a temperate question. A Zone IVb destination needs a different long-term condition.

19. How many batches are on ongoing stability, and how often is a batch added? One batch a year is a minimum. The answer should name the interval and the number of time points.

20. What happens to stability data when a change is made? Either the change is covered by existing data through a bracketing justification, or a new study starts. Silence means neither.

Block 5 — Documentation and regulatory (Q21–Q25)

21. Which documents accompany the first shipment? The nine-file pack is a reasonable baseline: certificate, SDS, specification, methods, stability summary, statement of origin, allergen and TSE declarations, packaging specification, and the quality agreement.

22. Is there a CEP or a DMF, and for which market? The answer is market-specific, and a certificate for one procedure does not open another. Details in the note on which confirmation applies where.

23. Who is the legal entity, and who signs the certificate? The signature should belong to a named function at a named site. A certificate signed by a trading entity tells you who sold it, not who made it.

24. How are deviations classified, and which classes are disclosed? Classification decides disclosure, and the customer should know the rule before an event rather than after.

25. What is the last audit outcome, including the findings? Not the certificate. The findings, and what was done about them.

Block 6 — Commercial and continuity (Q26–Q30)

26. What is the annual capacity for this product, and how much is committed? Capacity means nothing without the committed share. A site at ninety percent committed has no room for a new customer's growth.

27. What is the lead time from purchase order to shipment, and what drives it? The driver matters more than the number: a queue at the reactor behaves differently from a queue at the packing line.

28. Which single point of failure exists in the supply of this product? Every site has one. A supplier who names it has thought about it; one who denies it has not.

29. What happens if a batch fails after we have scheduled production? The answer describes the allocation rule under constraint, which is the rule that will apply in the year it matters.

30. Who is our technical contact, and what is their authority? A named person who can decide, rather than a shared inbox. The quality clauses should record the same name.

The three that decide the rest

Q2, Q6 and the dispute route implied by Q7. If the specification's origin is unknown and the method is unnamed, the other twenty-seven answers describe a product that cannot be compared with anything. If the precision is unknown, a disagreement between two laboratories has no resolution that is technical rather than commercial.

The rest of the list is valuable in proportion to those three. A supplier who answers them with documents will answer the others with documents. A supplier who answers them with adjectives will answer everything else the same way.

We publish this list because it is the one we are measured against, and because a customer who asks it of three suppliers will learn more in a week than a year of certificates can tell them. The frameworks behind the questions are ICH Q6A, Q7 and Q10, the Ph. Eur. texts and certificates, and the PIC/S guides.

The question worth asking yourself, before sending the list: if every answer came back as a document, with no covering note and no summary, would you know which three to read first, and would you trust the other twenty-seven without reading them?

Two follow-ups after the list

Answers are not verification. Two steps convert the list into a view of the supplier: ask for one complete document set for a single named batch, and ask for one reference customer whose material is used in an application similar to yours. The first tests whether the documents exist together. The second tests whether anyone else has relied on them.

Then wait ninety days and ask the same three questions from Block 2 again. Suppliers who answer consistently have a system. Suppliers whose answers drift have a person, and people leave.

One last note on how to read the answers. A supplier who answers twenty-eight of thirty questions with documents and admits to two gaps is in better shape than one who answers all thirty fluently. The fluent answer is a performance. The admitted gap is a fact about a system that knows its own edges, and a system that knows its edges can tell you where the risk is before you buy.

Frequently asked questions

How long should qualification take?

Long enough to answer the document questions and to see one batch's full record, which in practice is weeks rather than days. A qualification that completes in a day is a price comparison with a checklist attached.

Which three questions matter most?

Where the specification came from, what the method behind the critical numbers is, and what happens when a result differs between two laboratories. The first two determine whether a number means anything; the third determines whether a disagreement can be resolved without a commercial fight.

Should this list be sent to a supplier directly?

Yes, in two parts. Send the documentation block first, because it can be answered by email. Keep the audit block for the site visit, where the answers can be checked against records rather than described.

What if a supplier refuses to answer some of them?

Distinguish refusal from inability. A supplier who cannot answer a question about method precision has a documentation gap. A supplier who will not disclose the specification's origin has a positioning problem. The first is fixable; the second tells you what the relationship will be like.

Do small suppliers fail this list?

Not necessarily. Several questions are about documentation discipline, which is cheap, rather than scale, which is not. A small site with a maintained batch record and a named decision rule scores better than a large one that answers from memory.

Is there a scoring system?

Count how many answers are documents rather than statements. That single ratio predicts the rest, because a supplier who answers with records has the records, and one who answers with assurances is describing intent.