High-End Handbag Supplier Quality Audits Explained: The Decision Tree, the 68 Checkpoints and the 10-Day Correction Window

What does a high-end handbag quality audit actually check? It checks nine modules carrying 68 checkpoints, in a fixed order: documents and lot traceability first, then the bench, then the finished unit drawn at AQL 2.5 under ISO 2859-1. On a 500 pcs lot at general inspection level II that means 50 units, accept 3 and reject 4. Findings are scored into four bands, corrective actions close within 10 working days, and re-audit runs every 12 months with one unannounced window. Below: the decision tree, the evidence to request at each gate, and the records you should carry away.

auditor and atelier supervisor examining a stitched calfskin panel against a measurement sheet under a lamp — Gate 3 on the bench: a panel checked against a written tolerance rather than against an impression.

The Audit Decision Tree: Which Gate Comes First

A quality audit is not a single visit; it is a sequence of five gates, and each gate has to be passed before the next one is worth running. Run them out of order and you spend a day on the bench measuring a unit built from material nobody can trace, which is the most common way an audit wastes its own visit.

Gate 1, documents and lot traceability: can every claim be matched to a lot number and a date?
Gate 2, incoming material control: is the hide or fabric received, held and issued under a recorded reference?
Gate 3, the bench: are cutting, skiving, stitching and edge finishing done to a written tolerance?
Gate 4, the finished unit: do 50 drawn units from a 500 pcs lot pass at AQL 2.5?
Gate 5, close-out: is every finding tied to an operation, a cause, a fix and a date within 10 working days?

The order matters because each gate produces the evidence the next gate needs. Gate 2 gives you the lot number that Gate 4's test reports are issued against; Gate 3 gives you the operation names that Gate 5's corrective actions have to be written against. Start at Gate 4 and you will collect measurements you cannot interpret.

What an Audit Actually Measures

An audit measures three things only: whether a house can repeat a result, whether it can prove a material claim, and whether it can correct a fault at the station that produced it. Everything else on a checklist is a proxy for one of those three.

Repeatability is proven by records that exist before you arrive, not by a unit built for your visit. Provenance is proven by a lot number that resolves to a tannery and a test report. Correction is proven by a defect log written by operation, which is the only form a log can take if a fault is to be prevented rather than repaired. A house that shows you a beautiful unit but cannot produce those three artefacts has passed nothing.

Three claims, the evidence each one requires, and the appearance of failure. Note that all three forms of evidence are documents, not objects.
Claim	Evidence that proves it	What failure looks like
The result repeats	Measurement sheets from earlier lots, retained for at least 24 months	A sample built for the visit and no history behind it
The material is what is claimed	Tannery name, lot number, and test reports issued against that lot	A material description instead of a lot, and generic certificates
A fault can be corrected	A defect log written by operation with a cause named at each entry	A log written by unit, with faults repaired at the end of the line

The Nine Modules and the 68 Checkpoints

A programme audit runs nine modules carrying 68 checkpoints in total, scored out of 100, and takes one and a half to two days on site. The module weights are not equal, because the operations that cause expensive faults carry more checkpoints than the ones that cause visible but easily repaired ones.

1. Document control and traceability: 9 checkpoints; lot records retained 24 months
2. Incoming material and lot control: 8 checkpoints; shade banding under D65 light
3. Cutting and skiving: 7 checkpoints; panel tolerance ±0.5 mm, skive 0.6-0.7 mm
4. Panel preparation and interlining: 6 checkpoints; interlining 0.8 mm, no strike-through
5. Stitching and assembly: 11 checkpoints; 8-9 stitches per inch, Tex 40-60 thread
6. Edge finishing: 6 checkpoints; three coats, each flashed 8-10 minutes
7. Hardware and plating: 8 checkpoints; plating 0.15-0.3 µm, salt spray 24 h
8. Finished measurement and pairing: 7 checkpoints; overall tolerance ±1 mm
9. Packing, labelling and records: 6 checkpoints; carton drop and humidity control

The weighting is deliberate. Stitching and assembly carries eleven checkpoints because seam failure is the fault most often found by a customer rather than by inspection, and hardware carries eight because plating thickness cannot be judged by eye and is the least costly thing for a house to reduce without anyone noticing.

Sampling Mathematics: What AQL 2.5 Really Draws

AQL 2.5 under ISO 2859-1 does not mean 2.5 per cent of units may be defective; it means the sampling plan accepts a lot when the number of defects found in a drawn sample is at or below the accept number. On the programme minimum of 500 pcs at general inspection level II, the plan draws 50 units and accepts on 3 defects, rejecting on 4.

ISO 2859-1 single sampling plans, general inspection level II, AQL 2.5, normal inspection. The 500 pcs row is the one that governs a first capsule.
Lot size	Sample size	Accept	Reject
151-280 pcs	32 units	2	3
281-500 pcs	50 units	3	4
501-1,200 pcs	80 units	5	6
1,201-3,200 pcs	125 units	7	8
3,201-10,000 pcs	200 units	10	11

Two consequences follow. First, a small lot is inspected proportionally harder than a large one: 50 units out of 500 is ten per cent of the run, while 200 out of 10,000 is two per cent. Second, defects are counted against a written list agreed before bulk, which is why the defect list and the accept numbers have to be fixed in the same document as the price.

Material Evidence: Lot, Tannery and Test Reports

Material evidence is the gate that fails most often, and it fails in a specific way: the house can describe the leather but cannot name the lot. Because test reports are issued against a lot, an unnamed lot means the compliance work has no subject, and any substitution invalidates it.

The tannery name and the lot number behind the reference piece, not the type of leather.
Whether the tannery is audited under the Leather Working Group protocol, and at what rating.
ISO 3376 tensile strength and ISO 11644 finish adhesion, issued against that lot.
ISO 105 colour fastness, with ISO 11640 rub fastness at 500 dry and 100 wet cycles.
A per-lot declaration under REACH Annex XVII, including chromium VI at the 3 mg/kg threshold for leather.
California Prop 65 declarations where the goods are destined for the United States.
CITES documentation for any species listed in Appendix II or III, held before the hide is cut.

Leather thickness is part of the same evidence: a full-grain calfskin panel specified at 1.2-1.4 mm and skived to 0.6-0.7 mm at the turn-in is a specification, and a panel delivered at 0.9 mm is a different bag wearing the same name. Ask for the calliper reading on three units of the lot, not on the reference piece.

Hardware and Assembly Evidence on the Bench

Hardware is where a reduction is easiest to make and hardest to see, so the audit asks for measured values rather than appearances. Plating thickness, salt spray hours and cycle counts are the three numbers that separate a closure that will still work in three years from one that will not.

Plating thickness: 0.15-0.3 µm on brass, measured, not estimated
Salt spray proving: 24 hours minimum, 48 hours for hardware on a bag sold in a coastal market
Zipper: #5 coil or metal, run-tested over 100 open and close cycles
Magnetic closure: Pull force recorded, and the magnet seated so it cannot rotate
Stitch density: 8-9 stitches per inch, bonded nylon thread at Tex 40-60
Edge finish: Three coats, each flashed 8-10 minutes, edges burnished between coats
Interlining: 0.8 mm, applied so it does not strike through under the finish

A salt spray cabinet and a calliper on the bench are the two instruments whose presence tells you most about a house. Neither is expensive, and neither is bought by a house that does not intend to measure what it makes. Ask to see both used once, on your own unit.

brass turn-lock and zip pull on a bench beside a calliper and a salt spray test record sheet — Hardware evidence: plating thickness read with a calliper and proved over 24 hours of salt spray.

Scoring: Four Bands and What Each One Means

Sixty-eight checkpoints scored out of 100 resolve into four bands, and the band determines what you are allowed to do next. The useful discipline is to decide the band's consequence before the visit, so that a score cannot be argued upwards afterwards.

Four bands, their thresholds and the consequence each one carries. Decide the consequences before the visit, not after the score.
Band	Score	Meaning	Consequence
A	90-100	Records, tolerances and corrections all evidenced	Approved for bulk and for repeat seasons without re-visit
B	75-89	Sound, with two or three named gaps	Approved subject to corrective actions closing inside 10 working days
C	60-74	Structural gaps in records or in lot control	Conditional; a written plan and a second visit before bulk is released
D	Below 60	Evidence cannot be produced at all	Not approved; re-audit after remediation, no earlier than 90 days

Most first visits land in band B, which is a good result and not a disappointment. A house that has never been audited will always have gaps in document control, because documents are written for the auditor and nobody has asked before. What matters is whether the gaps close on time.

Corrective Action: The 10-Day Close-Out Rule

A finding without a close-out date is an opinion, and an opinion does not survive a season. Every finding has to be written as four fields: the operation that produced it, the cause at that operation, the fix, and the date, which is 10 working days from the report unless the fix requires new tooling.

Name the operation, not the unit: skiving station three, not bag number 14.
Name the cause: skive width drifted from 0.6 mm to 1.1 mm after a blade change.
Name the fix: blade change procedure now includes a first-piece check and a recorded reading.
Name the date: 10 working days, or the date new tooling arrives if tooling is genuinely required.
Verify by re-sampling 20 units from the corrected batch, not by inspecting the corrected unit.

The sentence that shows whether a finding is real

Ask the house to restate a finding in its own words. If it comes back as a promise to be careful, the cause has not been identified. If it comes back as a blade, a temperature or a sequence, the cause has been found and the fix will hold.

The fifth step is the one usually skipped, and it is the one that proves the fix. Re-sampling 20 units from a corrected batch at the same AQL is a two-hour exercise on 500 pcs, and it converts a written promise into a measurement you can file.

Re-Audit Cadence and Unannounced Windows

A full audit every 12 months is the standard cadence, with one unannounced window in the year for houses running more than two seasons for you. The unannounced visit is not a test of honesty; it is the only way to see the line as it behaves when nobody has prepared for you.

Scheduled annual

The full nine modules over one and a half to two days, with records prepared in advance and the bench available for questions. This is the visit that produces the score and the plan.

Unannounced window

A half-day visit covering modules 2, 5 and 6 only: material issue, stitching and edge finishing. Those three drift first, and they drift silently.

Triggered by change

Any new material family, a bespoke closure, or the addition of a second site should trigger a partial audit of the affected modules before bulk begins.

Triggered by a complaint

A field fault above two units in a delivery re-opens module 5 and module 8 regardless of the calendar, and the finding closes on the same 10-day rule.

Cadence costs little against the value of a season. Two days a year plus one half-day is under three days of attention on a programme that occupies 35-50 days of bulk production, and it is the only activity that tells you whether the other 47 days will behave.

Records You Should Leave With

Leave with documents, not impressions. Seven items are worth asking for before you leave the site, and a house that has run an audit before will have most of them ready.

The scored checklist, with every checkpoint marked pass, fail or not applicable.
The defect log written by operation, covering the last two completed lots.
The lot record for your material, including tannery, lot number and reservation status.
Test reports issued against that lot: ISO 3376, ISO 11644, ISO 105 and ISO 11640.
Per-lot declarations under REACH Annex XVII and California Prop 65.
The corrective action plan, with operation, cause, fix and date on each line.
The inspection plan for your lot: sample size, accept and reject numbers, defect list.

folder of audit documents with a scored checklist, lot records and test reports laid out on a table — The seven documents to leave with. Impressions do not survive a season; records do.

How Our Programme Prepares for an Audit

Our programme runs across 7 production lines with about 200,000 pcs per month of capacity and has held export experience since 2014, carried out through our validated partner factory Quanzhou Yingming Bags, an SGS-validated site. Sampling returns in 6-10 working days, 12-15 on a complex build, bulk runs 35-50 days, and inspection is drawn at AQL 2.5 under ISO 2859-1 with third-party inspection available where a buyer prefers it. Handover is on FOB Xiamen terms, and the swatch card set is billed at USD 4.80 and credited against the first order.

The sequencing detail sits under luxury sampling, the accept and reject arithmetic under luxury quality control, material provenance under luxury materials, and the full route on the high-end handbag sourcing page. A buyer evaluating a shortlisted house should also read high-end handbag sourcing for luxury buyers, which sets the sequence in which these checks are worth running.

Request the audit checklist before your next visit

Send your style count, destinations and the date you need goods delivered. You will receive the 68-checkpoint list, the inspection plan for your lot size and a calendar, within two working days. Request a quote.

Request a quote WhatsApp

Frequently Asked Questions

What does a high-end handbag quality audit actually check?

It checks nine modules carrying 68 checkpoints, scored out of 100, over one and a half to two days on site. The modules run from document control and lot traceability through cutting, stitching, edge finishing and hardware to packing records. Findings resolve into four bands, and each one is closed against an operation, a cause, a fix and a date.

How many units are inspected on a 500 pcs lot?

Fifty units. Under ISO 2859-1 at AQL 2.5 and general inspection level II, a lot of 281-500 pcs draws 50 units, with acceptance on 3 defects and rejection on 4. That is ten per cent of the run, which is proportionally far harder than the two per cent drawn on a 10,000 pcs lot.

Why does the audit start with documents rather than with the bags?

Because each gate produces the evidence the next one needs. Lot records give you the reference that test reports are issued against, and the bench records give you the operation names that corrective actions must be written against. Starting with the finished unit collects measurements you cannot interpret or act on.

What is the difference between a lot number and a material description?

A material description says full-grain calfskin; a lot number identifies the specific production batch at a named tannery. Test reports, REACH declarations and colour fastness results are issued against a lot, so only a lot number gives your compliance file a subject. A description cannot be substituted proof of anything.

Which test reports should be issued against the material lot?

At minimum ISO 3376 for tensile strength, ISO 11644 for finish adhesion, ISO 105 for colour fastness and ISO 11640 for rub fastness at 500 dry and 100 wet cycles. Add a per-lot declaration under REACH Annex XVII, including the chromium VI threshold of 3 mg/kg that applies to leather, and Prop 65 declarations for the United States.

What hardware measurements are worth asking for?

Plating thickness at 0.15-0.3 µm read with a calliper, salt spray proving at 24 hours and 48 hours for coastal markets, and a zipper run test over 100 open and close cycles. These three separate a closure that will still work in three years from one that will not, and none of them can be judged by eye.

How is the audit score turned into a decision?

Sixty-eight checkpoints scored out of 100 resolve into four bands. Band A at 90-100 approves the house for bulk and repeat seasons, band B at 75-89 approves subject to corrections closing in 10 working days, band C at 60-74 requires a written plan and a second visit, and band D below 60 defers re-audit by 90 days.

What makes a corrective action credible?

Four fields: the operation that produced the fault, the cause at that operation, the fix, and a date inside 10 working days. Ask the house to restate the finding in its own words. A promise to be more careful means the cause was never found; a blade, a temperature or a sequence means it was.

How often should an audit be repeated?

Every 12 months as a full nine-module visit, plus one unannounced half-day covering material issue, stitching and edge finishing, which are the three operations that drift first. Two further triggers apply: any new material family or bespoke closure, and any field fault above two units in one delivery.

What is an unannounced window actually for?

It is not a test of honesty. It is the only way to see the line as it behaves when nobody has prepared for the visit, since a scheduled audit always shows a bench at its best. Half a day on modules two, five and six is enough to reveal whether the written tolerances are the working tolerances.

Does AQL 2.5 mean 2.5 per cent of units may be defective?

No. It means the sampling plan accepts the lot when defects found in the drawn sample sit at or below the accept number. On 50 units drawn from 500 pcs the accept number is 3, which is six per cent of the sample and well below one per cent of the lot wrongly read as a defect rate.

What documents should I carry away from the visit?

Seven items: the scored checklist, the defect log written by operation, the lot record with tannery and reservation status, the test reports issued against that lot, the REACH and Prop 65 declarations, the corrective action plan with dates, and the inspection plan stating sample size, accept and reject numbers and the agreed defect list.

How long does a first audit delay a programme?

Rarely more than a few days if it is scheduled before bulk rather than during it. The nine modules take one and a half to two days, and corrective actions close in 10 working days. Set against sampling at 6-10 working days and bulk at 35-50 days, an audit is a small addition that protects the larger window.

Can third-party inspection replace an audit?

No, they answer different questions. Inspection judges a finished lot against a defect list at AQL 2.5; an audit judges whether the house can repeat a result, prove a material claim and correct a fault at its source. Third-party inspection is worth adding, particularly on a first capsule, but it reads output rather than capability.

Sources and further reading

Every figure on this page is either our own programme data or attributed below. Open the originals before you write them into a specification.

ISO — International Organization for StandardizationISO Publisher of ISO 2859-1, the sampling standard behind the AQL 2.5 plans and accept and reject numbers quoted here.
Leather Working GroupLeather Working Group Tannery audit protocol used to qualify a lot and to verify provenance claims during module two.
European Chemicals AgencyECHA Administrator of REACH, including the Annex XVII entry on chromium VI in leather that sets the 3 mg/kg threshold.
CITES — Convention on International Trade in Endangered SpeciesCITES Governs Appendix II and Appendix III documentation, which has to be held before an exotic hide is cut.
California Office of Environmental Health Hazard AssessmentOEHHA The authority behind California Proposition 65, which governs warning requirements for listed substances in goods sold in that state.