How to Read a Handbag QC Report and Act on Findings

A handbag QC report is decided by six numbers, and everything else in it is context: the lot size, the sample size, the AQL applied with its accept and reject figures, the defect count by class, the measured dimensions against tolerance, and the conclusion. The most misread line in any report is the conclusion, because a pass describes the sample and not the shipment: a lot that passes with five accepted major defects in an 80-unit sample still contains a proportional number elsewhere. Acting on findings means converting each count into one of five actions within 24 hours, ranked by the cost of delay: release, targeted sort, full re-sort, rework, or allowance. A re-sort costs 1-3 days and a re-make costs 35-50 days, so the decision is almost always to sort early rather than argue.

The six numbers

The Six Numbers That Decide Everything

Reports are long and mostly narrative, but the decision sits in six figures. Reading in this order takes about five minutes and produces the same conclusion every time, regardless of who wrote the report or how reassuring the accompanying email is. Everything else — the prose, the general comments, the photographs — is evidence for those six, not a substitute for them.

1. Lot size. What was presented, not what was ordered. A report on 800 pieces when 1,200 were ordered is a report on two thirds of the order, and the conclusion covers only that.
2. Sample size. Eighty units from a 501-1,200 piece lot at general inspection level II. If the sample is smaller than the standard requires, the accept numbers do not apply.
3. AQL with accept and reject. AQL 2.5 gives accept 5 and reject 6 for major defects at 80 units; AQL 4.0 for minor gives accept 7 and reject 8. Without both numbers the count means nothing.
4. Defects by class. Critical, major and minor counted separately. One critical fails regardless of the other two.
5. Measured dimensions. Actual values in millimetres against the specified tolerance, not ticks in a box.
6. Conclusion with a reason. Pass, hold or fail, and why. A pass without a stated reason is a summary, not a report.

Where two of the six are missing — usually the accept and reject numbers and the measured values — the document cannot be acted on. Ask for it to be reissued the same day rather than trying to reconstruct it, because a reconstructed report is not evidence if the decision is later challenged.

Reading the Header: What Makes a Report Valid

Before reading a single defect, check four fields in the header, because a report that fails any of them is not usable regardless of what it found. These checks take a minute and they are the difference between a document and a piece of paper.

• Order reference and style. Does the report describe the order you placed, including the colourway? A report that says black does not cover tan.
• Inspection type and date. Pre-shipment, final or inline, with the date, because each type supports a different decision.
• Inspector name and body. Traceability matters if the finding is disputed months later, particularly where a third party such as SGS ran the inspection.
• Reference artefact. Which approved sample or specification revision the inspection measured against, with its date. Measuring against a superseded revision invalidates the result.

The last item is the one most often wrong. A specification revised after sampling — a tolerance widened, a stitch density changed — creates two competing references, and an inspection measured against the older one will report defects that the newer revision permits. Keep one dated revision in force, referenced by the report, and archive the rest.

The Defect Table: Counts, Classes and Cost

The defect table is where the money is. Each row is a count of observations in one class, and the useful reading is not the total but the distribution: how many types, and how concentrated. Three majors of one type is a single cause with one fix; three majors of three unrelated types is normal variation.

Concentration is the signal worth acting on. Where 4 of 5 majors are the same type, the cause is one operation and the fix is a bench correction plus a targeted sort, which is cheap. Where the 5 majors are 5 different types, the cause is the general workmanship level, and the only real remedies are a full sort, an allowance or accepting the returns.

The Measurement Table and What Tolerance Means

Measurement is the part of the report most often reduced to a tick, and it is the part a buyer most needs in numbers. A table that says within tolerance three times tells you nothing you can compare; a table that records 268 mm against a specified 265 mm plus or minus 3 mm tells you the run is sitting at the top of its tolerance and will drift out of it if the jig slips.

Actual values

Millimetres, per dimension, on at least three units

Specified value and tolerance

Stated alongside the actual, not implied

Units measured

Identified by carton so a pattern can be traced

Strap drop

Measured unloaded from the top edge to the strap apex

Instrument

Steel rule or tape, stated, since a cloth tape stretches

Out-of-tolerance units

Listed individually, not averaged away

Averages hide the problem. Three units measuring 262, 265 and 280 mm average to 269 mm and look acceptable against a 265 mm target, while the third unit is 15 mm out and will be returned. Require individual values, and require the out-of-tolerance units to be listed separately rather than folded into a mean. The same principle governs the tolerances used in inspection.

Judging the Photographs

Photographs are evidence only when they are identifiable, scaled and lit. A good report photograph has three properties, and a set that lacks them cannot support a decision or a later claim, however many images it contains.

• A scale or a reference in frame. A ruler, a coin or the bag itself, so the viewer can judge size. A close-up with no reference is the single most common weakness in an inspection report.
• The defect in context. One wide shot showing where on the bag the defect sits, plus one close shot of the defect itself.
• Consistent lighting. Defects photographed under the same source as the colour evaluation, since colour appearance changes with the light.
• A caption tying it to the count. Each photograph referenced to a row in the defect table, so the count and the image agree.

Where a report contains twenty photographs and four defects, be suspicious: image volume is sometimes used in place of analysis. What matters is one photograph per defect class, tied to a count, plus the measurement table. Anything beyond that is useful for the record but does not change the decision.

Turning Findings Into One of Five Actions

Every finding resolves to one of five actions, and the correct one is determined by the class of defect and by how much of the run is still loose. Decide within 24 hours, because the cost of every option rises with each day the cartons sit packed and the vessel booking approaches.

1. Release. Defects at or below the accept numbers, no criticals, and measurements inside tolerance. Record the types for the next run.
2. Targeted sort. The defects trace to identifiable cartons or to a known band of the run. Cheapest effective option at 1-2 days.
3. Full re-sort. Defects spread across the lot, or one critical found. Costs 1-3 days plus a re-inspection fee.
4. Rework. The defect is repairable, such as a thread end, a mis-set rivet or a light adhesive mark. Hours per hundred units, followed by re-inspection.
5. Allowance. Cosmetic defects where the goods can still be sold, at 2-5 percent of invoice value, with the terms in writing.

The ranking is by the cost of delay rather than by the nominal cost, which is why sorting beats arguing. A day spent debating whether a defect is major is a day in which the vessel booking tightens, and the difference between a 1-2 day targeted sort and a 35-50 day re-make is the difference between a late launch and a missed season.

What a Report Does Not Tell You

Four things are outside the scope of even a good report, and knowing the limits is what prevents a buyer from over-trusting a pass. The report describes a sample drawn on one day; it does not describe the whole lot, the durability of the product, the compliance position or the units that were never presented.

• It does not cover unpresented goods. Where the report covers 800 of 1,200 pieces, the remaining 400 are uninspected.
• It does not test durability. Workmanship is visible; a seam that fails at three months is a laboratory question, covered by cycle and pull tests.
• It does not establish compliance. Restricted substances are a laboratory matter under REACH, Proposition 65 and CPSIA, not an inspection observation.
• It does not predict the second order. Only a retained specification, retained tooling and retained units do that.

Programmes coordinated through our validated partner factory Quanzhou Yingming Bags, with export experience since 2014, issue the report to the buyer before the balance moves and pair it with laboratory documentation where the market requires it. That combination is what makes the handbag quality control reference actionable rather than theoretical.

Building a Lot-by-Lot Record

One report is a decision; a series of reports is a trend. Keep a folder per style containing the specification with its revision date, every inspection report, the laboratory reports, the packing lists and the retained-unit log. After three lots the series will show whether the programme is improving, and it will usually show it before anyone notices anecdotally.

• Log defects per unit by class. Total defects divided by units inspected, split by class; a controlled programme sits near 0.35.
• Note the top three defect types per lot, with the action taken and whether it worked.
• Track the measured dimensions, so a drift toward the edge of tolerance is visible before it becomes a failure.
• Record the action and its cost, which is what makes the next decision faster.
• Keep two retained units per lot for 90 days, referenced in the same folder.

That record is also what shortens the second order. Where the specification is unchanged and the series shows stability, sampling can often be skipped entirely, removing 6-10 working days from the calendar, and the repeat can run straight to bulk at 35-50 days with inspection at AQL 2.5 under ISO 2859-1.

What the Inspection Costs, Booked by Man-Day

Inspection is priced in man-days rather than in units, which is why it behaves like a fixed cost on a small lot and disappears per piece on a large one. The figures below assume typical southern China day rates and single sampling at general inspection level II. Two additions are worth asking about in advance: travel beyond the city, usually USD 40-90 per visit, and re-inspection, commonly charged at 60-80 per cent of the original fee. Set against a re-sort at USD 0.35-0.60 per piece, the arithmetic favours inspecting a small lot earlier rather than later.

Switching Rules: Normal, Tightened and Reduced

A single lot's result is not the whole system. ISO 2859-1 moves between three states according to recent history, so two reports with identical defect counts can sit in different regimes and carry different consequences. Buyers who understand switching can see coming changes in cost and can also use it deliberately: a programme that has earned reduced inspection spends its quality budget on process control instead, which is the cheaper place to spend it.

The commercial consequences are measurable. Reduced sampling removes roughly a third of the man-days across a multi-lot programme, which on four lots of 1,200 pieces is about USD 350-450 a season. Tightened sampling does the opposite deliberately: it raises the probability that a marginal lot fails, which is the point, since the alternative is discovering the same three lots of drift in a customer returns queue at USD 22-28 per unit instead.

Turning a Count Into a Rate You Can Compare Across Lots

Raw defect counts cannot be compared between lots of different sizes, so every count should be converted before it goes into the folder. Defects per unit, and then defects per hundred units per operation, are the two readings that travel. A controlled programme sits near 0.35 defects per unit overall; what matters more is whether the figure is concentrated in one operation, because a concentrated figure has one cause and one fix.

Record the same two numbers every lot and the series becomes a management tool rather than a filing habit. Augment the total with the count per operation, which is what tells you whether a correction is holding across a season, and note the days lost to each action so the next decision is made against cost rather than against memory. Three lots is enough to see a direction; six is enough to argue about capacity with evidence.

Frequently Asked Questions

People Also Ask