THE QUICK ANSWER
Write a tolerance beside every harness dimension before the next order. For each size, name the measurement point, the state the product is measured in, the target, and the permitted variation in each direction. Then agree in the purchase order how many pieces are measured, by whom, and what happens to a reading outside the band. Our development sheet already asks for target, unit and tolerance on every dimension, because a target alone cannot be inspected.
A target without a tolerance cannot be inspected
A tolerance is the range inside which the goods conform. Until it is written, an inspector holding a tape has nothing to compare a reading to, and a piece that measures differently from the approved sample cannot be called conforming or non-conforming. Our development sheet for a custom dog harness lists six dimensions to fill in — neck opening minimum and maximum, chest girth minimum and maximum, back panel length, webbing width, finished weight and packed carton volume — and the approval record for each one is target, unit and tolerance. We publish no values in those fields. The size run belongs to your range, not ours. What the sheet fixes is the shape of the answer: every dimension carries three things, the number you want, the unit, and how far the goods may sit from it. Fill the third field at the same time as the first two, on every line and for every size.
Fix the measurement point and the state before you argue about the number
A dimension is only reproducible once the method is written. Our sample approval guidance already makes the point for soft goods: agree how the item is positioned and measured, because compression and handling can change the result. So each line in the table needs one sentence covering three things. The landmark, meaning where the tape starts and finishes. The state, meaning closed on a form or laid flat, and which adjuster position. And the handling, meaning whether tension is applied and how much. One worked line, which you can paste into a specification: chest girth is measured with the harness closed on the size form, the girth adjuster set to its marked middle position, the tape following the panel seam, no tension added. Then photograph the measurement being taken and attach that photograph to the specification revision, so the method travels with the number rather than with the person who wrote it.
The size chart measures the dog. The specification measures the goods.
These are two different documents and they are easy to confuse. AKC guidance identifies chest girth and neck girth as key harness measurements and notes that brand sizing varies. Ruffwear's measuring guide starts with girth around the widest part of the chest behind the front legs. Both describe how to measure an animal. Neither describes the product you are buying. Your published chart says which dogs a size is for. Your specification says which pieces the factory may ship. The two are joined at one point: the shortest piece the tolerance permits must still reach the top of the range you published. So set the maximum-adjustment target above the published maximum by at least the negative tolerance, not level with it. The decision that follows is a publishing decision. Either derive the chart from the specification's worst permitted case, or hold the chart until the first run has been measured. Changing a live size chart after launch is a separate problem, and it touches every listing that references it.
Write the tolerance per size, and write it asymmetric
Over target and under target do not cost the same. A chest girth above target is absorbed by the adjuster. The same amount below target removes the top of the published range for that size. So the two directions are separate decisions, not one number with a plus-or-minus in front of it. As an illustration only — we publish no harness size run, and these figures are not a standard — a 48 cm chest girth target might be written +1.5 / -0.5 cm rather than ±1.0 cm, because the risk falls on one side. Then check the band against your own grading before you issue it. A ±1.5 cm tolerance describes a 3 cm band. If consecutive sizes are graded 6 cm apart, that band is half the distance to the next size, and the bands of adjacent sizes only meet once the tolerance reaches ±3 cm. Run that arithmetic per size. A band that is comfortable at the top of the run can consume most of the interval at the bottom.
If you cannot set the number, ask the factory to propose it and state the basis
If you have no tolerance history to draw on, that is not a reason to leave the column blank. Put the question into the RFQ instead: for each dimension in the specification, propose a tolerance, state the measurement method it assumes, and state which sizes it applies to. Then read the answer for what it contains rather than for what it promises. A usable answer names the dimension, names the state the piece is measured in, gives a figure in each direction, and says which sizes share it. A weak answer says standard tolerance, or gives one plus-or-minus figure for the whole chart with no method attached. If the reply arrives in an email, ask for it as a revision of the specification instead. Whatever is agreed applies to that product and that order, and it belongs in the written quotation, because it varies with product, size run and customization.
Name the sample size in the order rather than assuming one is in force
A tolerance is only useful if enough pieces are measured to see the run rather than one piece. There is no default here to inherit. Our commercial terms list inspection method, sample size, defect classes, acceptance criteria, report format and release authority as things agreed before production, which means the sample size is a term you write, not a standard already running in the background. Write three things into the booking. How many pieces are measured and where they are drawn from, separate cartons across the shipment rather than the top of the first pallet. That the readings are recorded as numbers, one piece per row, with the carton reference beside each. And who books and pays for the inspection. Two figures then come off that sheet: the average against target, and the distance between the lowest and highest reading. An on-target average with a wide spread and a run sitting consistently under target are different findings and get answered in different places.
Put it in the order, because a claim is assessed against the specification
A tolerance agreed by email is not a term of the order. Put the measurement table in the specification, give the specification a revision number, and reference that revision in the purchase order. Then agree what evidence a claim needs and what the remedy process is, in writing, before production. Our published position on claims is narrow and worth reading literally: keep the purchase order, SKU, batch reference, inspection record, photographs and the quantity affected; rework, replacement, credit, refund or another remedy is assessed against the agreed specification and the documented evidence. No refund or credit policy is assumed, and nothing on this page is an offer of one. Which remedies apply, in what order, and who carries the cost of the inspection that decides, are confirmed in the written quotation and purchase agreement for your product and your order. Ask for that paragraph at RFQ stage, while it is still a negotiation rather than a dispute.
A useful comparison
| Dimension on the development sheet | State it is measured in | Tolerance written against | What a reading outside the band changes |
|---|---|---|---|
| Neck opening (min / max) | Flat and closed, inside edge to inside edge | Each size, both ends of the range | The size no longer matches the opening published on your chart |
| Chest girth (min / max) | Closed on the size form, adjuster at its marked middle position | Each size, minimum and maximum separately | One end of the published range for that size can no longer be met |
| Back panel length | Flat, top seam to bottom seam on the centre line | Each size | The size no longer sits where the grading placed it |
| Webbing width | Flat and relaxed, across the strap | The material, per approved reference | The strap no longer matches the approved reference |
| Finished weight | A single piece, unpacked | Each size | The freight assumptions were built on a different weight |
| Packed carton volume | Sealed export carton, as packed | The pack-out, not the piece | The freight quotation no longer describes the shipment |
Explore private-label product options.
No-pull dog harness — materials, customization and sample checks Standard dog leash — materials, customization and sample checks Adjustable webbing collar — materials, customization and sample checksSources and research context
Sources reviewed 2026-09-14. Community excerpts are anecdotes. Manufacturer guidance applies to that manufacturer’s products. Shortlists are editorial comparisons without hands-on comparative testing or verified sales rankings.
- American Kennel Club · How to measure a dog for clothes and gear
- Ruffwear · How to Measure for Your Dog’s Gear
TURN THE ANSWER INTO AN OEM BRIEF
