Drawn from our own bag line and the audit records buyers ask to see before a first order.
The carton label said 45 by 13.8 in, the polybag label said 45 by 13.8 in, and the bag inside measured 43 by 13 in. A buyer in Rotterdam found the mismatch on the third carton of a pre-shipment check, with 1,800 more bags already wrapped and waiting for a container. The factory's QC file for that order was four pages long. None of the four pages recorded a single measured dimension.
That buyer’s problem was not a factory without quality control. It was a factory with quality control nobody could read. Over the following weeks she sent five questions to her supplier, one at a time, and each answer changed what she asked next. Her sequence is worth copying, because it moves from claims to documents to the one gap that almost every file leaves open.
The question that arrives before the first purchase order

Her first message was short: what do you actually check, and when? The reply listed incoming material inspection, in-process monitoring, finished product inspection, and pre-shipment sampling. Four stages, correctly named, and completely unverifiable on its own.
A factory that runs all four stages will have traces of each one. Incoming inspection leaves resin certificates and roll receipts. In-process work leaves logs at the loom and the sewing station. Finished goods inspection leaves a report with sample sizes and defect counts. Pre-shipment sampling leaves a signed sheet tied to a carton range.
Ask for one trace per stage before you send a deposit. A supplier who answers with a brochure has described a system. A supplier who answers with four dated sheets has shown one. The difference takes ten minutes to establish and saves months of argument later.
Does the factory test the bag the way the buyer's customer will use it?

Her second question came from a returns pattern she had seen before: bags that passed inspection and failed in a moving truck. The reason sits in the difference between static and dynamic load. A static test hangs a filled bag and waits. A dynamic test drops it, stacks it, and drags it, which is what a moving crew does all day.
Static ratings flatter a bag. Dynamic capacity can run well below the static figure, so a handle that holds a hanging load may still tear when the same bag is dropped from height with weight inside. A factory that only hangs bags has tested the calmest moment of the product’s life.
Ask specifically for a drop test record: drop height, load inside, number of repeats, and which bags were pulled from the batch. Then ask where failures show up. On woven bags the answer should be handles and stitch lines, because that is where dynamic force concentrates. If the factory reports failures somewhere else, the test probably was not a drop test.
Which three documents prove the checks actually happened?

By the third exchange she had stopped asking about systems and started asking for paper. Three documents carry most of the weight.
The first is the raw material Certificate of Analysis for the polypropylene resin, with the Melt Flow Index stated. For PP resin the working range is 2 to 4 g/10min. Above it the bag turns brittle, below it the bag goes soft and loses shape. A generic certificate without an MFI figure tells you the resin exists, not that it suits your bag.
The second is the in-process QC log. Look for weave temperature held between 160 and 180°C for PP, and stitch count per inch recorded per batch. Heavy duty bags should hold a minimum of 8 stitches per inch. A log with these two columns filled in for every batch is evidence of a habit. A log with the columns empty is evidence of a template.
The third is the final AQL report, with critical, major, and minor defects separated and the sample size stated. Seam failure and handle tear-out belong in the critical column. Color mismatch and loose stitching belong in major. A report that lumps everything into one defect count cannot tell you whether the order is safe to ship.
One practical filter applies to all three: ask for them within one business day. A factory with digital records sends them quickly. A factory that needs a week is reconstructing them, and reconstructed records tend to be tidy.
Why does the same spec produce two different bags?

Her fourth question came after two approved samples from the same supplier felt different in the hand. Same fabric weight on paper, same dimensions, different stiffness and a different drape. The spec sheet had not changed. The inputs had.
Weave density is the usual culprit. Woven PP should sit at 10 to 12 picks per inch in both warp and weft, and a mill that runs the loom slightly faster can land inside the tolerance on average while leaving thin bands across the roll. Those bands rupture first under load. Tension unevenness does not show in a photo and rarely shows in a hand feel until the bag is filled.
Thread is the second culprit. A factory under cost pressure may keep high strength thread on the top spool and switch the bobbin thread to something weaker. The seam looks correct from outside and fails from inside. A stitch tensile record, run to ASTM D5034, is the document that catches this, and it should be batched rather than issued once for the whole year.
Ask for the pick glass readings from five random samples per roll, and ask for the bobbin thread specification in writing. Neither request is unusual. A supplier who treats them as unusual has told you something useful.
What should a pre-shipment sample inspection look at first?

Her fifth question was about sequence, because the Rotterdam cartons had been checked in the wrong order. The inspector measured the print and the color first, ran out of time, and never opened a tape measure on the body.
Dimensions come first, because a size error invalidates everything downstream. Measure the flat width and height on bags pulled from at least three carton positions, and compare against the approved sample, not against the label. Labels get copied forward from order to order, and a copied label is how a 16.9 in bag ships inside a 17.7 in carton.
Stitch density and handle attachment come second. Count stitches per inch at the handle junction and along the side seam, and pull the handle by hand before any gauge is involved. A handle that shifts under a firm hand pull will not survive a loaded drop.
Print and color come third, checked against the approved swatch under the same light the customer will see. Colorfastness testing to AATCC 16, at 20 hours of xenon arc exposure, covers the warehouse sunlight question, with an acceptable delta E below 3.0 against the original. That test belongs to the material file, not to the pre-shipment day, but the swatch comparison belongs to both.
Packaging comes last and gets its own pass. Carton label against polybag label against measured bag, all three in one line of the report. The Rotterdam mismatch would have surfaced in the first ten minutes under this order.
How does a buyer read a test report without a lab background?
She admitted she could not evaluate the tensile numbers herself, so she asked a simpler question: which line on this page would change if the factory had cut a corner? That reframing works better than learning the standards.
On a tensile report run to ASTM D5034, look for warp and weft reported separately, with the fabric weight stated alongside. For 200 GSM fabric the floor sits at 800 N warp and 700 N weft. A report that gives one combined figure has hidden the direction that usually fails.
On a bursting strength report to ASTM D3786, look for the fabric diaphragm result against the 450 kPa minimum. This number matters most if your customer stacks loaded bags on a concrete floor, where puncture and pressure do the damage.
On a colorfastness report, look for the exposure hours and the delta E. Twenty hours of xenon arc stands in for roughly three months of warehouse light. A delta E under 3.0 is the pass line.
On a handle pull test, look for the load held, the hold time, and the stitch count at the junction. A handle rated for a 50 kg bag should hold double that for a minute without tearing, and the junction should carry at least 8 stitches per inch. If the report omits the stitch count, the pull result describes a sample, not a production method.
The question buyers ask after the pallet has already shipped
The Rotterdam buyer eventually got her replacement cartons and her dimension record. What she did not get, and had never thought to request, was the batch traceability link: which resin lot, which loom run, and which sewing line produced the bags in carton 47.
That question only becomes urgent when a defect appears in the field and the factory needs to isolate the affected run. Without a lot number on the carton and a matching entry in the in-process log, every bag in the shipment becomes a suspect, and the recall scope becomes the whole order. With the link, a single loom run can be quarantined and the rest released.
Ask for it before the pallet moves. One carton label field, one log column, and one sentence in the QC file that connects them. It costs the factory almost nothing to add and costs a buyer an entire container to do without.
The five questions she sent, in order, were: what do you check and when, do you test the way the buyer’s customer uses it, which documents prove it, why do two identical specs feel different, and what does the pre-shipment check look at first. Each one is answerable with a sheet of paper. The sixth question, about traceability, is the one that decides how small a problem stays.
FAQ
What are the four types of quality inspection in a bag factory?
Incoming material inspection, in-process monitoring, finished product inspection, and pre-shipment sampling. Each stage should leave its own record, and a buyer can request one trace per stage before placing a first order.
How many picks per inch should a woven PP bag have?
Woven PP bags should hold 10 to 12 picks per inch in both warp and weft. Uneven tension creates thin bands that rupture under dynamic load, so readings should come from several random samples per roll rather than one spot check.
What stitch count per inch should a heavy duty bag hold?
A minimum of 8 stitches per inch at the handle junction and along load-bearing seams. The count belongs in the in-process log per batch, and it should appear on any handle pull test report next to the load and hold time.
Can a factory send QC documents within one business day?
A factory with digital records can. The three core documents are the resin Certificate of Analysis with the MFI stated, the in-process log with weave temperature and stitch count, and the final AQL report. A week long delay usually means the records are being assembled after the request.

