How to Handle Size Tolerances in Fabric Bag Manufacturing

Why fabric cannot hold metal tolerances

Creators coming from hardware or electronics bring a tolerance habit that does not transfer to sewn goods. An injection-moulded part holds a tenth of a millimetre because the tool determines the dimension. A fabric panel does not: it is cut from a material that stretches on the bias, sewn with a seam allowance that varies with operator pressure, and then pressed, stuffed and folded. Each stage adds movement, and the movements accumulate rather than averaging out.

The practical consequence is that a fabric bag dimension is a nominal target with a distribution around it, and the distribution is the thing to specify. Asking for a bag to be 30.0 cm wide is asking for something no sewing line can deliver consistently. Asking for 30.0 cm plus or minus 5 mm, measured flat at a stated point after pressing, is a specification that can be quoted, sampled, inspected and argued about on evidence.

This matters commercially because dimensional complaints are among the most common post-dispatch issues on bag rewards, and most of them concern units that are inside any reasonable tolerance. A backer who measures a bag at 29.6 cm against a page that says 30.0 cm will report a fault even though the unit is correct. The fix is partly in production and largely in how the dimension is communicated. The crowdfunding bag guide covers the wider specification path.

Minimum order quantity

500 pcs per style / colourway

Sampling

6-10 working days (12-15 for complex builds)

Bulk production

35-50 days after deposit + approved pre-production sample

Inspection

AQL 2.5 to ISO 2859-1

Trade term

FOB Xiamen

Payment

T/T 30/70

Production floor

4,950 m² SGS-verified, 137 staff, 7 lines, 149 machines

Monthly capacity

200,000 pcs per month

Programme contact

service@junyuanbags.com

Where the variation actually comes from

Five stages add movement, and knowing which stage dominates tells you where a tolerance reduction is realistic. Cutting introduces the first variance, because fabric shifts under the blade and under the marker. Sewing introduces the second and usually the largest, because seam allowance depends on how the operator feeds the panel. Pressing and finishing introduce the third. Stuffing and packing introduce the fourth. And the measurement itself introduces a fifth, which is why the measurement method belongs in the specification.

Two of those rows explain why a tolerance cannot simply be tightened by asking. Seam allowance variance is a human factor: two operators on the same line produce different results, and the spread is inherent to feeding fabric through a machine. That is why a tolerance reduction on a sewn dimension usually requires a jig or a folding fixture, which is tooling, which is cost and lead time. Our bag manufacturing page sets out where fixtures enter the sequence.

Realistic tolerance bands by construction

The band you can hold depends on how many times the fabric is handled and how much structure is built into the panel. A flat tote cut from two panels and sewn on three sides is the most controllable object in the range. A structured backpack with piping, binding, a frame sheet and a padded back panel is the least, because every one of those elements adds a seam and a fold.

Notice that the tightenable column never reaches zero and never gets close. A campaign that needs a tighter fit than the right-hand column allows is not asking for a better bag; it is asking for a different construction, usually with a rigid insert or a moulded element, and that is a design decision with its own cost and lead time. Decide it at specification stage rather than discovering it at inspection.

Writing tolerances into the tech pack

A tolerance clause has four parts, and most packs have one. Writing all four is what converts a dimension from a wish into a specification that an inspector can apply and that a production partner can quote against. The four parts are the nominal value, the accepted band, the measurement condition, and the measurement point.

1. Nominal value: the target dimension, stated in centimetres to one decimal place.
2. Accepted band: the plus or minus tolerance that defines a passing unit, stated in millimetres.
3. Measurement condition: whether the unit is measured flat, empty, pressed, and after how long at ambient.
4. Measurement point: where the rule is placed, referenced to a stated seam, edge or centre line.
5. Instrument: a steel rule or tape, with the tension stated where a tape is used.
6. Reference unit: the approved pre-production sample, against which any dispute is resolved.

The measurement point is the part most often omitted and the one that causes most arguments. A bag measured across the top opening and the same bag measured across the base can differ by more than the tolerance band, simply because the panel is not parallel. State the point: across the base seam, at the centre of the front panel, or between two stated handles. Then photograph the measurement on the approved sample so there is no ambiguity later.

Interfaces that must actually fit

Most dimensions on a bag can carry a generous tolerance because nothing depends on them. A few cannot, because something has to fit inside them: a laptop sleeve, a bottle pocket, a packing cube that nests into a travel bag, or a strap that has to reach a stated length. These are the dimensions worth tightening, and they are worth tightening because a failure there makes the unit unusable rather than merely imprecise.

• A laptop sleeve should be specified by the device size class it accepts, not only by internal dimension.
• Any sleeve needs clearance: an internal width stated to fit a 14-inch device should exceed the device by 10-15 mm.
• Nested components need a stated clearance between the inner and outer part, not two independent tolerances.
• Strap length should be specified as an adjustable range with the minimum and maximum measured under no load.
• Zip openings should be stated by usable opening width, not by overall bag width.
• Test every fitted interface with the physical object it must accept, at sampling, and record the result.

Measuring and recording at sampling

Sampling is where the tolerance clause becomes real, because it is the only point at which a dimensional problem can be corrected on one unit instead of five hundred. Measure the sample against every stated dimension, record the actual values rather than pass or fail, and photograph the measurement at the stated point. Actual values matter because they tell you where in the band the process is sitting.

1. Measure every stated dimension and record the actual value in millimetres, not a tick.
2. Photograph each measurement with the rule visible and the stated measurement point identifiable.
3. Measure two or three units, not one, so the spread is visible rather than a single point.
4. Test every fitted interface with the physical object at the stated clearance.
5. Compare measured values against the nominal and note whether the process is centred in the band.
6. Record the results on the sample approval, so the bulk run has a numeric reference, not only a physical one.

Measuring two or three units is the step that reveals a process problem. If a nominal 30.0 cm dimension comes back at 29.6, 30.1 and 30.4 cm, the process is centred and inside band, and the run will be fine. If it comes back at 30.4, 30.5 and 30.5 cm, the process is drifting to one edge of the band, and a proportion of the bulk run will exceed it even though every sampled unit passes. That is the single most useful number in the whole exercise. Our sample prototyping page sets out where measurement sits in the checkpoint sequence.

Bulk drift and how it happens

A run that is inside tolerance at first-off can drift out of it by unit four hundred. The causes are mundane: a different operator on a different shift, a new roll of fabric with slightly different stretch, a press parameter moved to keep up with output, or a marker re-nested after a cutting error. None of these is a defect in itself; all of them change dimensions.

The control that catches all six is a mid-run dimensional check. Ask for five units to be pulled partway through the run, measured against the same stated points, and photographed beside the approved pre-production sample. It takes an hour of line time and it converts a potential five-hundred-unit problem into a correction at unit two hundred and fifty.

Dimensional defects in an AQL plan

Pre-shipment inspection to AQL 2.5 under ISO 2859-1 draws a statistically determined sample and applies accept and reject thresholds to the defects found. Dimensional defects have to be classified like any other, and the classification determines whether a unit that measures 6 mm over on a plus or minus 5 mm specification fails the run. That is a decision to make before inspection day, not on it.

• Define dimensional tolerance as a major defect where it affects fit, and as minor where it does not.
• State the measurement method in the inspection work instruction, matching the tech pack clause exactly.
• Require the inspector to measure the same points stated in the pack, against the approved reference unit.
• Set the sample size and accept and reject numbers under ISO 2859-1 general inspection level II.
• Decide whether the inspector measures empty, flat and pressed units, and say so in the instruction.
• Require the report to list measured values, not only defect counts, so drift is visible.

Asking for measured values rather than counts is the most useful single change most campaigns can make to their inspection instruction. Two runs can both pass at AQL 2.5 with identical defect counts, and one can be centred in the band while the other sits at its edge with a tail outside it. Only recorded values show which is which. Our QA and QC page sets out how the defect classes are structured.

Stating size to backers honestly

A campaign page that states 30.0 x 38.0 x 12.0 cm invites a backer to measure and find a discrepancy. A page that states 30 x 38 x 12 cm nominal, or that states a size class such as fits a 14-inch laptop, invites a comparison against a real object and is far more likely to satisfy. This is not about hiding anything; it is about stating the number at the precision the manufacturing process can support.

The same principle applies to capacity. A soft bag's litre figure is a calculation from dimensions that are themselves approximate, and a claim of 18.3 litres implies a precision that the object does not have. Say approximately 18 litres, or better, show a photograph of the bag next to objects a backer recognises. Photographs settle the question that numbers raise.

When to tighten a tolerance and what it costs

Tightening a tolerance is possible within limits, and it is worth doing on the two or three dimensions where fit matters. It is not worth doing across the board, because a blanket tightening raises cost, raises the reject rate, and produces a bag that is no better for the backer. The mechanism is always the same: a tighter band needs a fixture, a slower operation, or a higher reject allowance, and all three cost money.

Note that the tooling route has a calendar consequence as well as a cost one. A fixture has to be made and trialled, which is one of the reasons a complex build moves the sampling window from 6-10 to 12-15 working days. Where a tight tolerance genuinely matters, build that into the plan from the start rather than asking for it after the first sample comes back.

Shrinkage and movement in service

Tolerance is not only a manufacturing question; it is also an in-service one. Natural fibre bags move when washed, coated fabrics can relax in heat, and padded structures compress under load and recover imperfectly. A bag measured at goods-in and measured again after a backer washes it will not be the same object, and the specification should anticipate that rather than be surprised by it.

• State expected shrinkage for natural fibre fabrics, measured after one wash, as a percentage.
• Wash one sample during the sampling round and re-measure it before approving the specification.
• Specify relaxation time before packing, so units are not measured or packed while still compressed.
• Choose lining and interlining with compatible shrinkage, or the panel will distort after washing.
• State a care instruction that matches the tested behaviour, not an aspirational one.
• Where shape retention matters, specify the interlining grade and test recovery after compression.

A tolerance clause you can reuse

The clause below can be dropped into any bag tech pack. It is deliberately explicit about condition, point and instrument, because those three are what make the clause enforceable. Adjust the band to the construction using the table earlier in this article, and keep the reference to the approved sample, which is what resolves any disagreement.

• External dimensions are nominal: width 30.0 cm, height 38.0 cm, depth 12.0 cm.
• Accepted tolerance: plus or minus 5 mm on all external dimensions, unless stated otherwise below.
• Measured flat, empty, unloaded, after pressing, at ambient temperature, 24 hours after finishing.
• Measured with a steel rule placed at the stated point: across the base seam for width, centre front for height.
• Laptop compartment: accepts a 14-inch device with a minimum 12 mm clearance on width and depth.
• Any dimension outside the stated band is a major defect under AQL 2.5 to ISO 2859-1.
• The approved pre-production sample is the reference unit for any dimensional dispute.

With that clause in the pack, the quotation can be accurate, the sample can be measured against something, the inspection has a method to apply, and the campaign page has a defensible way to state the size. That is the whole of what tolerance work is for: not to make the bag perfect, which is impossible, but to make the agreement about what good enough means explicit before five hundred units are sewn. The custom bag categories page and the laptop bag page show how the clause is applied to specific formats.

• Crowdfunding bag sample prototyping
• Kickstarter campaign bag QA and QC
• Bag manufacturing for campaign creators
• Custom bag categories
• Custom laptop bags for Kickstarter
• Custom travel bags for Kickstarter
• Material customisation for Kickstarter bags
• Custom bags for Kickstarter campaigns

Frequently Asked Questions

People Also Ask