The Waterproof Bag Supplier Audit Checklist That Predicts Real Output

Why process records outrank certificates on the wall

Most buyers arrive with a supplier audit checklist that is really a document-collection exercise, and then treat the signed paperwork as the finding. It is the wrong order of evidence. A certificate proves that an auditor once visited and that a fee was paid; a process record proves what the line was doing last Thursday when nobody was watching. When you are choosing a waterproof bag supplier for a programme that will run for years, the second is worth far more than the first, because waterproof performance is produced by parameters — temperature, pressure, dwell, cooling — and parameters leave traces.

There is a structural reason for this. Certifications are purchased artefacts, sampled and scheduled, with a known visit date and a preparable evidence set. Process records are by-products: they exist because someone had to write a number down every two hours to do their job, and they accumulate whether or not anyone ever reads them. That means the record set is both harder to fabricate at scale and much more revealing when read as a series rather than as individual entries.

The practical consequence is that your audit day should be organised around records, with the floor walk as a cross-check against them. If the welding log says 165 degrees Celsius every two hours for three months and the thermocouple calibration record was last updated eleven months ago, you have learned something no certificate would have told you. If the leak-test sheet shows zero failures across four thousand units and the rework bin is full of seam repairs, you have learned something else entirely.

None of this means certificates are worthless. They are a floor, not a finding: useful for confirming that a management system exists and that a recognised body was willing to attach its name to it. The ISO 9001 quality management framework is the usual starting point, and the audit technique itself is standardised in the ISO guidance on management system auditing published by the International Organization for Standardization. What they cannot do is tell you whether this specific line, on this specific shift, is currently in control.

The pre-audit packet: what to demand two weeks out

The audit begins before you travel. Send a written request for a specific document set roughly two weeks ahead, and make the request specific enough that it cannot be answered with a brochure. The quality of the response is itself data: a supplier who can produce three months of weld parameter logs in two days has them; one who needs to "prepare" them for a week is telling you something about how the line is actually run.

Two behavioural tells matter more than any single document. First, does the supplier ask why you want a particular record? A factory with nothing to hide sends it; one that asks whether the order is confirmed before sending it is running a cost-benefit calculation on transparency. Second, do the documents arrive as scans of working sheets or as freshly typeset summaries? Working sheets have handwriting, crossings-out, coffee stains and sequential serial numbers. Summaries have been rewritten for you, and rewriting is where inconvenient rows disappear.

Finally, ask for the same packet in the same form every time you audit. Consistency over years is what turns an audit from a snapshot into a trend, and a supplier who knows you will ask for the weld logs again next spring has a standing incentive to keep them honest.

Walking the floor in process order, not in the order offered

The default factory tour is a route optimised for the supplier, not for you: reception, meeting room, the tidiest line, the packing area, then lunch. Insist instead on walking in process order and in reverse at least once. Start at dispatch and walk backwards to goods-in, or start at goods-in and follow the material through. The reason is simple: material flow reveals where work sits, and work-in-progress piles are the most honest indicator of where the process is constrained or where rework is quietly accumulating.

• Ask to see the material store before the line, not after; unlabelled rolls and open part-used laminate are where substitution begins.
• Follow one identified batch from cutting through welding to packing and count how many times it is handled and recorded.
• Stop at every buffer between stations and ask how long work normally sits there; a large buffer usually hides a mismatch in station output.
• Look for the rework area deliberately. Every factory has one; the question is whether it is visible, counted, and fed back into the process.
• Finish in the packing area and ask to see a carton from an order other than yours being closed.

While walking, watch the stations rather than the product. At a high-frequency welding station the things worth seeing are whether the operator has a reference sample within reach, whether parameters are displayed and readable from where they stand, whether there is a gauge or timer in use, and whether the first unit of a batch is separated and checked. Those four observations take ninety seconds and tell you more about process discipline than an hour in the meeting room. Our page on high-frequency welding in waterproof bag manufacturing sets out what a controlled station looks like in detail.

Also note what is absent. No reject bin at a station means rejects are either not being made, which is implausible, or being removed informally, which is worse. No reference sample means the operator is working from memory. No parameter display means the machine was set once and is trusted indefinitely, which is precisely the assumption that produces a run of weak welds three months later.

Ask the operator, not the supervisor

Supervisors answer questions about the process as it is documented. Operators answer questions about the process as it is performed, and the gap between those two answers is one of the highest-yield findings available to you. Do it respectfully and without an audience: step to the operator, ask an operational question, and do not let the supervisor answer for them.

The questions should be concrete and non-accusatory. What do you do if the machine temperature alarm goes off? What do you check on the first bag of a new batch? How do you know a seam is good? What happens to a bag you think is bad? Who signs the sheet, and when? Each of these has a documented answer and a performed answer, and where they diverge you have found the real process.

Two answers are worth listening for specifically. "I would tell my supervisor" is a correct answer that also tells you the operator has no authority to stop the line — so the containment depends on one person being present. "I would stop and adjust" tells you something different and usually better, provided the adjustment is recorded. And an operator who cannot articulate what a good seam looks like, or who points to a supervisor across the room as the definition, is working without a criterion.

Do this at two or three stations and compare. Consistency of answer across operators means the process is genuinely understood and trained. Divergence means the documented process and the working process have separated, which is the normal state of most lines and the reason audits should sample people rather than documents.

Three months of records, not today’s form

A single day of records can be produced to order. Three months cannot be fabricated convincingly, because fabricating a series requires inventing variation, corrections, absences and shift differences, and almost nobody does that well. This is why the most useful sentence in an audit is "show me the last three months", delivered calmly and then waited out in silence.

Read the series, not the values. A weld temperature log that sits at exactly 165.0 for ninety consecutive days is not evidence of control; real processes wander within a band, and a band with no wander has been typed rather than measured. A peel test series that trends gently downward over three weeks and then jumps back up is not noise either — it is usually a thermocouple replacement or a material lot change, and either is worth asking about. This is the same logic that underpins statistical process control: the shape of the series is the information.

Then check the gaps. Missing weekend entries are normal. Missing entries on the Wednesday before a public holiday, or a run of identical entries that begins exactly on the day a new QC hire started, are not. Gaps are where the real story usually sits, and a factory that can explain its gaps credibly has a live system; one that cannot has a decorative one.

Ask for the records to be produced from the working file rather than from an archive folder prepared for visitors. Working files are organised by date and are messy. Audit folders are organised by document type and are tidy. Tidiness, in this specific context, is evidence of preparation rather than of control.

Four checks specific to welded waterproof construction

Generic factory audits miss the category entirely, because they check stitching, packing and housekeeping and never touch the thing that makes the product work. A waterproof programme lives or dies on the bond, so the audit has to inspect the bond.

The leak test deserves particular attention because it is the check most often performed for the visitor and skipped for the customer. Ask to see it run, on a unit from a lot that is not yours, at the depth and duration stated in the specification. If the apparatus is present but dusty, or if the method described verbally differs from the written procedure, the programme has a waterproof claim rather than a waterproof control.

Cross-check against the acceptance framework you intend to buy under. A factory that understands critical defects will tell you unprompted that a leak has a zero acceptance number and will have set the leak-test sample size by risk rather than by a cosmetic sampling table. Our AQL sampling guide and the complete inspection guide give you the vocabulary to test that understanding in ten minutes.

Seven tells that you are seeing a prepared audit

Prepared audits are not dishonest by definition; some preparation is professional courtesy. The problem is that preparation converts the visit from a measurement into a performance, and a performance cannot be extrapolated. Learn to see the seams.

• Every document you ask for is produced within minutes, from a single labelled folder, in a consistent new typeface.
• The floor is recently swept to an unusual standard, but machine interiors, filters and jigs are greasy and untouched.
• Operators glance at a supervisor before answering, or the supervisor physically steps in to answer for them.
• The line is running your product, or nothing at all, rather than the mix of orders a real week would contain.
• Records exist for the current week in detail and for prior months only as monthly summaries.
• Welding parameters are displayed but the log sheet shows no entry since the start of the day.
• The reject and rework area is empty, or is described as "only for minor issues" without a count attached.

The corrective move for each is the same: move from the artefact to the trace. If the folder is suspiciously complete, ask for the source file on the computer that produces it. If the floor is clean, ask to see the machine maintenance log. If the line is running nothing, ask what ran yesterday and to see yesterday’s output record. A real operation has traces everywhere; a staged one has a surface.

It is also fair to say this out loud. A sentence like "I would rather see an ordinary day than a prepared one, and I will not penalise normal mess" genuinely changes behaviour at some suppliers, and the ones it does not change are the ones you most needed to identify.

Timing the visit: shift, day and season change the answer

The same factory audited on a Tuesday morning and on a Saturday night is two different factories. Night and weekend shifts typically run with thinner supervision, less maintenance cover and more temporary labour, and they are where quality systems degrade first. If a meaningful share of your volume will be made on those shifts, audit them.

Day of week matters too. Monday morning is where the weekend’s parameter drift and unmerged material lots surface; Friday afternoon is where schedule pressure and early release produce shortcuts. Peak season is worse and more informative: a factory running at capacity with overtime and subcontracted operations is under exactly the stress that will be applied to your order when it matters most.

The strongest version of this is the semi-announced visit: give a window rather than a date — "some week in the next two months" — so the supplier cannot stage a single day but can still arrange access. Fully unannounced audits are rarely practical in this category and can damage a working relationship; the window approach gets most of the value at a fraction of the friction.

Whichever you choose, write the timing into the agreement. A contractual right to audit on reasonable notice, including a right to inspect during production rather than only before shipment, is worth asking for at the same time as you negotiate the rest of the purchase agreement, because a right you did not secure is a favour you will have to ask for later. The same negotiating session is where you agree inspection access, so treat the two together rather than discovering the gap later — our manufacturer negotiation guide covers how to raise it without poisoning the discussion.

Social and environmental compliance is a different audit

Buyers often fold labour and environmental compliance into the quality audit and then conclude that a clean quality result means a clean compliance result. They are different systems with different evidence, different auditors and different failure modes, and conflating them is how brands end up surprised. Working hours, wage records, age verification, health and safety, and waste handling each have their own documentary trail, and in several jurisdictions your obligations as an importer are separate from anything the factory promises.

The practical approach is to run them separately and to use a recognised framework rather than an improvised one. Social compliance programmes such as those administered by amfori provide standardised protocols precisely because improvised social audits miss things predictably. Our own page on BSCI audit compliance covers what buyers should expect from that side of the file.

One overlap is worth keeping: audit fatigue. If a factory is being audited weekly by different customers on different protocols, its real state is whatever it does when nobody is there, and more audits will not reveal it. Ask how many audits they hosted last year and by whom. A factory that hosts twelve and shares findings openly is usually in better shape than one that hosts two and treats each as an exam.

Grading the report: four tiers with consequences attached

An audit report that ends in a narrative paragraph is useless, because nobody can act on it and nobody can track it. Grade the result, and attach a decision to each grade before you audit, so that the grade drives behaviour rather than a discussion.

Two rules make the grading work. First, separate critical from major from minor using the same defect classes you will later buy under, so that an audit finding and an inspection finding use one language — the classification set out in our quality assurance guide transfers directly. Second, treat fabricated records as an automatic D regardless of what the floor looks like, because a supplier who will fabricate a weld log will fabricate an inspection report, and no amount of subsequent inspection can compensate for that.

Write the grade to the supplier. A shared grade with stated consequences is a management tool for them as much as for you — it gives the quality manager internal leverage to get the thermocouple calibrated. A grade you keep private is just a note in your own file.

Corrective action closure: the part everyone skips

Audits produce findings; findings produce nothing unless they are closed. The closure loop is mundane and it is where nearly all audit value is lost. A finding without an owner, a date and a verification method is a wish, and programmes that skip verification end up believing their fixes rather than knowing them.

• Write the finding as an observation with evidence, not as a conclusion: state what you saw, where, and which record supports it.
• Require a root cause rather than a remedy — "thermocouple replaced" is a remedy; "no calibration schedule existed" is a root cause.
• Assign a named owner at the supplier with a date, and a named owner on your side who will chase it.
• Set the verification method at the same time: what document, what measurement, or what re-observation will prove it.
• Verify on the next production run rather than on a prepared demonstration, and record the result against the original finding.

Watch for the two standard evasions. The first is "corrected", meaning the single instance was fixed while the condition remains — one thermocouple was replaced rather than a calibration schedule created. The second is "training provided", which is the least verifiable corrective action in existence and should always be paired with a system change, because training without a system change expires with the next staff turnover.

Keep a simple register: finding, owner, date, evidence of closure, date verified, verified by whom. It takes an afternoon to maintain and it converts your audit programme from a series of events into a trend line, which is what lets you answer the only question that matters about a supplier — are they getting better or worse?

Why a passed audit does not mean stable quality

This is the core judgement and it deserves to be stated plainly: an audit passing tells you the factory was capable on one day under conditions it controlled. It does not tell you that the next lot will be good, and treating a passed audit as a quality guarantee is one of the most expensive mistakes in this category. Capability and stability are different properties, and only the second one protects your brand.

The reasons are structural rather than cynical. Capability is demonstrated once under favourable conditions; stability has to survive material lot changes, staff turnover, schedule compression, a second shift, a machine moved to a different bay, and a supplier substitution two tiers down that nobody mentioned. An audit is a photograph of capability. Stability is a film, and only repeated data produces it.

That is why the audit result should change your inspection strategy, not end it. A supplier graded B gets tighter incoming and pre-shipment control until a run of clean lots earns reduced inspection; the switching rules that make sampling plans responsive exist precisely for this. Our pre-shipment inspection checklist is the instrument that runs after the audit, not instead of it.

The right mental model is that the audit buys you the right to place the first order with eyes open, and every subsequent lot buys you data. Suppliers who understand this welcome the inspection, because consistent data is how they earn reduced inspection and better terms. Suppliers who resent it are telling you that their audit result was the product, not their process.

What to do when the audit fails, and when to walk

Not every failed audit ends the relationship, and not every one should. The distinction is between a gap that can be closed and a posture that cannot. A calibration schedule that does not exist can be created in a month. A supplier who produced fabricated records, or who refused access to a process area, has told you how they will behave when something goes wrong on your order, and that is not a gap — it is a disposition.

Ask three questions before deciding. Did they disclose the problem themselves at any point, or did you find all of it? Do their explanations match the documents, or contradict them? And did they propose a system change or a promise? System changes indicate a factory that can improve; promises indicate one that hopes the problem will not recur while you are watching.

If you continue, continue conditionally: smaller first order, tighter inspection, defined re-audit date, and a written agreement on what happens if the re-audit fails. If you leave, leave cleanly and say why — a specific, evidenced explanation is both fairer and, in a small industry, more useful than silence. Our wider sourcing guide covers the qualification sequence that sits around this decision.

And if you are rebuilding a supply base, remember what the audit was actually for: not a score, but a picture of whether this factory tells you the truth about its own process. Minimum order quantity on our waterproof programme is 500 pieces per style, sampling takes 6–10 working days with test evidence attached, bulk runs 35–50 days, and quotations are issued FOB Xiamen. The full chain from first enquiry through sampling into bulk, including which records you can ask to see at each gate, is set out in the production process guide on our main site.

Frequently Asked Questions

Q1. What is the single most important thing to check in a waterproof bag factory audit?

Three months of welding parameter logs and peel-test results. They are the hardest records to fabricate convincingly, and unlike certificates they show whether the bond that makes the product waterproof is actually under control.

Q2. How long should a supplier audit take?

A meaningful first audit is a full working day on site, plus half a day beforehand reading the pre-audit document packet. Anything shorter becomes a tour, and a tour produces photographs rather than findings.

Q3. Should I announce the audit in advance?

Give a window rather than a date — for example, some week in the next two months. The supplier can arrange access but cannot stage a single day, which is where most audit value is lost.

Q4. Why is asking operators better than asking supervisors?

Supervisors describe the documented process; operators describe the performed one. Where the two diverge is the real process, and that gap is one of the highest-yield findings available in a day on site.

Q5. What does a fabricated record look like?

Identical values to the decimal across every entry, no handwritten corrections, all calibration dates in the same week, and detailed current-week records backed only by monthly summaries for prior months.

Q6. Do certificates like ISO 9001 matter at all?

They are a floor, not a finding. They confirm a management system exists and a recognised body was willing to certify it, but they cannot tell you whether this line on this shift is currently in control.

Q7. What are the four checks specific to waterproof construction?

Weld parameter control, destructive seam testing on real coupons, the leak test method and sample size, and seam tape or adhesive control including shelf life. Generic audits miss all four.

Q8. How do I know whether the leak test is real or performed for visitors?

Ask to see it run on a unit from a lot that is not yours, at the stated depth and duration. Dusty apparatus or a verbal method that differs from the written procedure means the claim exists but the control does not.

Q9. Should I audit the night shift?

Yes, if any meaningful share of your volume will be made there. Night and weekend shifts run with thinner supervision and more temporary labour, and that is where quality systems degrade first.

Q10. How should I grade an audit result?

Use four tiers with consequences committed in advance: approved, approved with conditions, probationary, and not approved. Fabricated records or refusal of access should be an automatic not-approved regardless of how the floor looks.

Q11. What makes a corrective action credible?

A root cause rather than a remedy, a named owner with a date, and verification on the next production run. "Training provided" alone is the least verifiable corrective action and should always be paired with a system change.

Q12. Is a passed audit enough to guarantee good quality?

No. An audit is a photograph of capability on one prepared day. Stability has to survive material lot changes, staff turnover, schedule compression and shift changes, and only repeated lot data demonstrates it.

Q13. What should happen to my inspection plan after an audit?

The audit should change it, not end it. A supplier with conditions gets tighter incoming and pre-shipment control until a run of clean lots earns reduced inspection under the normal switching rules.

Q14. Should social compliance be audited at the same time?

Run it separately. Different evidence, different protocols and different obligations. Ask how many audits the factory hosted last year — heavy audit fatigue with openly shared findings is usually healthier than two audits treated as exams.

Q15. When should I walk away from a supplier after a failed audit?

When they fabricated records, refused access, or explained problems in ways that contradict their own documents. Those are dispositions, not gaps. Disclosed, document-consistent gaps with system-level fixes are usually worth a conditional second chance.

Q16. How often should I re-audit an approved supplier?

Every eighteen to twenty-four months for a grade A supplier, annually for one approved with conditions, and within six months for probationary. Re-audit immediately after any critical field failure regardless of schedule.

Q17. Can I audit during production rather than only before shipment?

You should secure that right contractually when you negotiate the purchase agreement. A right you did not secure becomes a favour you have to ask for at exactly the moment you need it most.

People Also Ask

What should a waterproof bag supplier audit checklist include?

A pre-audit document packet, a floor walk in process order, three months of process records, the four waterproof-specific checks, operator interviews and a graded report.

How do you tell a prepared audit from a real one?

Look for traces rather than surfaces: source files, machine maintenance logs, yesterday’s output records and a non-empty rework area with counts attached.

Are ISO certificates enough to qualify a bag factory?

No. They establish that a system exists, not that the line is in control. Process records and floor observation are what establish the second.

What is a critical finding in a factory audit?

Anything that affects waterproof function, safety or legal compliance, plus fabricated records or refusal of access. Critical findings block new programmes until closed.

How often should buyers audit suppliers?

Every eighteen to twenty-four months for strong performers, annually where conditions apply, and immediately after any critical field failure.

Does passing an audit guarantee quality?

No. It shows capability on one day. Stability across material lots, shifts and schedule pressure is demonstrated only by repeated lot data and ongoing inspection.