Custom Pet Carrier Sample Approval Steps

Pet carrier production desk · Updated 2026-10-07 · 14 min read

For a custom cat carrier the ventilation question is where the mesh sits, not how much there is: high side panels plus a shaded front zone keeps airflow without visual exposure.

Approval is a sequence of decisions, not a single yes

A sample arrives, it looks broadly right, and the brand approves it. Three weeks later the bulk shipment has a zipper that catches, a logo placed 8 mm higher than the mock-up, and a baseboard that flexes. None of those three defects was present in the photograph the brand approved, and all three were present in the physical sample that somebody did not open.

That pattern repeats across nearly every programme we see, and the cause is structural rather than careless. A pet carrier is a soft assembly of twenty to forty components — shell panels, lining, foam, baseboard, mesh, webbing, binding, zippers, sliders, buckles, D-rings, feet, labels, thread — and a single physical sample cannot evidence all of them. Some properties are visual and can be approved from a photograph. Others, such as foam recovery, zipper travel under load, or whether a baseboard creases when the unit is lifted by one handle, can only be assessed by handling the object.

Splitting approval into gates solves this by matching the evidence type to the decision type. Colour is approved on a lab dip, not on a finished bag. Structure is approved on a prototype, not on a drawing. Dimensional conformance is approved against a spec sheet on a confirmation sample. Compliance is approved on a test report. Each gate tests the thing it can actually test.

The second benefit is that gates make cost visible. A change requested at the brief gate is free. The same change requested at prototype costs a rebuild. Requested after the sealed reference, it costs a container. Buyers who can see that gradient make different decisions, and programmes run better because of it.

There is a psychological element too. A single approval event invites a holistic verdict — "it feels about right" — and holistic verdicts are where unwritten expectations survive. Named gates force specific answers to specific questions, and unwritten expectations are exactly what surfaces later as a dispute.

Gate zero: the brief has to be signed before anything moves

Gate zero is documentation, and it is the gate most often skipped because it produces nothing physical. It is also the gate where a programme is cheapest to change.

A workable brief carries: the size ladder with weight bands, the target retail price band, the destination markets and therefore the compliance set, the shell and lining constructions, the hardware finish, the logo artwork with its placement measured from a stated datum, the colourway list with Pantone or physical references, the packing format, and the target delivery date. Eight items. A brief missing any of them will generate a question later, and every question later costs more than the same question now.

The two items that cause the most trouble are logo placement and compliance set. Logo placement specified as "centred on the front panel" is not a specification, because panels are curved and stretch; placement specified as "centred horizontally on the front panel, 45 mm below the top binding, on a 250 mm wide placement area" is. Compliance set matters because testing requirements follow from destination: a programme selling into the EU and the US needs a different evidence pack from one selling into a single market, and the testing cannot be retrofitted after production.

Our production team returns a technical response to the brief within the sample window, flagging anything that cannot be made as drawn. That response is the real output of gate zero — a written list of what will change and why, agreed before cutting starts. A structured brief template is the fastest way to get there, and the brand-side identity pieces covered in the brand kit note belong in the same package.

Gate one: colour and trim approved before a panel is cut

Colour approval happens on swatches, and it happens first. This is the gate that protects the dye lot, and it is the gate brands most want to compress.

A lab dip is a small piece of the actual production cloth dyed to the target reference. It is usually produced in two or three variants around the target, because a dyed shade sits differently on a 600D polyester than it does on a paper Pantone chip. Approving from the chip rather than the dip is the single most reliable way to get a colour that is technically correct and commercially wrong.

The assessment conditions matter more than most buyers expect. Dyed textile shades are evaluated under controlled lighting, because a metameric pair — two colours that match under one light source and not another — will pass in a showroom and fail in a store. The standard approach is to view under at least two specified illuminants, typically a daylight source and a store-lighting source, and to record the decision against both.

Tolerance is expressed numerically. A colour difference is measured as a delta E value against the approved standard, and a working commercial tolerance for a dyed-to-order shell is around 1.0 to 1.5 delta E for adjacent production, with looser tolerance on textured or coated constructions. Stating the number at this gate is what makes the second order match the first, and the topic is worked through in detail in our note on the reorder process.

Trim is approved on the same gate: webbing, binding tape, zipper tape, thread, mesh and hardware finish, each as a physical swatch attached to a card that is signed and retained. Hardware finish in particular should be approved against a physical sample, because plating appearance varies with base material and batch. Where a programme specifies a corrosion requirement, that is evidenced later at the testing gate, commonly by a salt-spray method such as ASTM B117 for salt spray testing.

Gate two: the prototype is allowed to be wrong in specific ways

The prototype answers one question: does the thing work as an object? It is not expected to be finished, and treating it as finished is a mistake in both directions — buyers either approve too early or reject for cosmetic reasons that were never in scope.

What the prototype is meant to establish is structural. Can the pet be put in and taken out without the opening collapsing. Does the baseboard hold its shape when the unit is lifted by one handle with a realistic load in it. Do the shoulder straps sit where a person's shoulders are. Does the mesh panel stay open under tension or does it pull the shell into a curve. Does the unit stand upright on a floor when loaded.

What the prototype is explicitly not meant to establish is cosmetic finish. Stitch consistency, logo position, binding alignment and colour match are all expected to be approximate at this stage, because the pattern is still being corrected. A brand that rejects a prototype for 3 mm of logo drift will pay for an extra build cycle to fix a problem that would have been resolved by the pattern correction anyway.

The right way to run the gate is a handling session with a weight in the bag. Put 6 kg in a medium carrier, carry it across a room, set it down, lift it by each handle in turn, open and close each entry, and see what moves. Photograph everything. The output of the gate is a numbered corrections list, written against those photographs, with each item marked as structural or cosmetic.

Structural items trigger a pattern change and a rebuild. Cosmetic items are carried forward to the confirmation sample and fixed there. Separating the two is the whole point of the gate, and it is what keeps a programme to one rebuild instead of three.

Gate three: the confirmation sample and the corrections list

The confirmation sample is the first unit that is expected to be right. Every structural correction from the prototype gate has been applied, trim is from the approved swatch card, and the logo is placed to the brief. This is the gate where the programme is either approved or starts over.

The discipline here is the corrections list. Each numbered item from the prototype gate is checked off individually against the new sample, with a photograph, and either closed or re-opened. Approving a confirmation sample without walking the list is how items silently survive into bulk. A list with twelve items and no closure record is a list that will produce a dispute.

Dimensional conformance is checked at this gate and nowhere else. Every dimension on the brief — external length, width, height, opening dimensions, strap length range, pocket dimensions, and packed flat dimensions — is measured on the confirmation sample and recorded on a spec sheet alongside the approved value and the tolerance. The tolerance band is normally plus or minus 5 mm on external dimensions for a soft carrier, and tighter on openings where a pet's fit is involved.

The table below sets out the full gate sequence with the working days and the deliverable each gate produces, on the standard 6-10 working day sample window.

Note that gates overlap rather than run strictly end to end. Test evidence is commissioned during the confirmation window so the results arrive with the sample rather than after it, which is why the physical build fits 6-10 working days while the full approval sequence occupies roughly three weeks of calendar time. Confusing the two is the commonest cause of a missed launch date.

Gate four: test evidence arrives as a gate, not an attachment

Compliance testing is frequently treated as paperwork assembled after approval, and that ordering is backwards. Test results can change the design, which means they have to arrive while change is still cheap.

The evidence set depends on destination market. A programme selling into the United States carries general product safety obligations administered by the Consumer Product Safety Commission, plus state-level chemical marking requirements. A programme selling into the European Union carries the chemicals framework described by ECHA under REACH, and in practice retailers ask for restricted-substance declarations on every component, not just the shell.

Mechanical testing is programme-specific and usually includes a static load test on the handles and shoulder straps, a baseboard deflection test, a zipper cycle test, and a drop or impact test on the loaded unit. Where the product carries a safety claim, independent evaluation matters: organisations such as the Center for Pet Safety publish protocols for pet travel products, and aligning an internal test to a published protocol is far more defensible than an in-house method with no external reference.

The operational rule is that testing is commissioned against the confirmation sample, not the prototype. Testing a prototype produces a report on an object that no longer exists. Testing the confirmation sample produces a report on the object that bulk will replicate, which is the only useful result.

Two practical points about commissioning. Book laboratory slots before the sample exists, because third-party capacity is the most unpredictable duration in the whole sequence and is booked weeks ahead at certain points in the year. And submit enough units: most mechanical and chemical protocols consume several samples destructively, and a laboratory that receives one unit will return a partial report and start the clock again.

Gate five: the sealed reference that bulk is judged against

The last gate produces the object that settles every later argument: a pre-production sample, made on the production line with production tooling and production materials, signed and dated by both parties, sealed, and retained.

Two copies is the minimum, one held by the brand and one by the production side. Where a programme runs across multiple partner facilities, each holds one. The sample is the reference for the bulk inspection at AQL 2.5, for any dispute about colour or construction, and for the reorder twelve months later. Without it, every one of those conversations becomes a negotiation about memory.

The pre-production sample is also where the packing specification is fixed. Carton dimensions, units per carton, pallet pattern, carton marks and label placement are all confirmed on a real packed unit rather than on a calculation. This is the point at which a packing error is cheap to fix and the point after which it is not.

Sealing should be literal. A tag through a tamper-evident seal, signed across the join, dated, photographed, and the photographs stored with the order record. It sounds excessive and it takes two minutes. It is also the only thing that makes a later claim about what was approved resolvable in an afternoon rather than a week.

The pre-production unit is also the moment to confirm that the line can actually hold the specification at speed. A sample built slowly by a senior operator proves the design; a pre-production unit built at production pace proves the process. Differences between the two — stitch density dropping, binding tension varying, logo placement drifting — are precisely what the pre-production round is for.

Once the seal exists, the 35-50 day bulk clock starts. Nothing before it starts the clock, and a brand that believes production began when the prototype was approved will misjudge its landing date by three weeks.

Who signs, and what a signature actually commits

Every gate needs a named signatory, and the name matters more than most brands expect. An approval returned by "the team" is an approval nobody owns, and it is the reason disputes about what was agreed end up unresolved.

The clean split is by competence rather than seniority. Colour and trim are signed by whoever owns the brand's visual identity, because a delta E tolerance is a design decision with a commercial consequence. Structure and dimensions are signed by whoever owns product, because a strap geometry change is a functional decision. Compliance evidence is signed by whoever carries regulatory risk. The sealed reference is signed by both parties, because it is the only artefact that binds both sides.

A signature should also commit to what was not said. The standard form our production team uses records three fields: what is approved, what is explicitly not approved, and what remains open. That third field is the useful one. A brand that approves a sample while noting "logo position to be confirmed" has not approved the sample, and recording it that way prevents the confirmation from being read as blanket.

Revision control follows from signing. Each approval carries a revision letter, and any subsequent change increments it. Bulk is made to the highest revision that has been sealed. This sounds bureaucratic and takes almost no time, but it is the mechanism that stops a superseded drawing reaching a cutting table — which is otherwise a common and expensive accident.

Finally, the record itself. Approvals, corrections lists, dimension sheets, swatch cards and photographs belong in one folder per programme, indexed by gate, accessible to both sides. A programme with a complete record can be audited, reordered from, and handed to a new buyer without a single phone call. A programme without one cannot.

Four ways the sequence breaks, and how to recognise each

Every failed approval sequence we have reviewed fails in one of four ways, and each has a recognisable early signal.

Scope creep during approval. A change that arrives mid-sequence — a different lining, an extra pocket, a revised logo size — resets the gate it belongs to. The signal is a corrections list that grows between rounds rather than shrinking. The fix is contractual rather than technical: changes after gate three are quoted as a new sample cycle with its own dates.

Artwork arriving late. Logo artwork is the most common late input, and it blocks the confirmation sample because embroidery and print placement cannot be verified without final files. The signal is a sample requested before vector artwork exists. The fix is to make artwork a gate-zero deliverable, alongside the brand identity kit.

Unapproved substitution. A component is unavailable and something similar is used, and nobody says so. This is the failure mode that produces the worst disputes, because the sample the brand approved is not the sample bulk replicates. The signal is a component on the sample that does not appear on the swatch card. The fix is simple and worth insisting on: any substitution is declared in writing and re-approved before bulk.

Approval by photograph. The sample is assessed from images rather than handled. Foam density, zipper travel, baseboard stiffness and strap comfort cannot be judged from a photograph, and a programme approved this way will discover its structural problems on the container. The signal is an approval returned within an hour of the courier's delivery scan.

Underlying all four is the same discipline: every gate produces a written artefact signed by a named person. Our production team maintains that record under a quality system audited to ISO 9001, with BSCI social compliance across the SGS-verified production base, which is what makes the record auditable rather than anecdotal. QUANZHOU JUNYUAN BAGS has run pet carrier programmes since 2014, its founder having been in the trade since 2004, from a 4,950 square metre base with 137 people, 7 production lines and 149 machines.