Pet Bag QA: Quality Assurance System for Wholesale

Wholesale pet bag sourcing desk · Updated 2026-10-06 · 14 min read

Pet bags oem china and pet bags odm china arrangements differ on who owns the tooling, and Market & Business Strategy is usually where that argument starts. Pet bag bulk supplier terms should state mould ownership in writing before the first sample is cut.

QA as a System, Not a Department

Quality assurance is often treated as a team that sits beside production and checks its work. That is quality control. Assurance is the set of decisions made before and around production that determine whether the work is checkable at all.

The distinction is practical rather than academic. A specification exists before an inspector needs one; a procedure exists before an operator needs one; an approval exists before a bulk run starts. Assurance is the work done in advance so that control has something to check against.

The system view is what makes it valuable. Assurance is a set of linked processes, not a list of documents, and the links are what matter: the specification feeds the procedure, the procedure feeds the training, the training feeds the operation, and the audit checks the whole chain.

Ownership is the element most often missing. Every element of an assurance system needs a named owner who is accountable for its currency, and a document without an owner becomes out of date within a season.

The buyer's interest in this is direct and commercial. A programme with functioning assurance produces repeatable results, which means a reorder behaves like the first order, a specification change is implemented correctly, and a defect does not recur.

The evidence of a functioning system is unglamorous: current documents with revision dates, approvals with signatures, records completed at the time rather than retrospectively, and corrective actions with verification. That is what an audit looks for.

The evidence of a decorative system is equally recognisable: documents with no revision history, records completed in the same handwriting for an entire season, and corrective actions that describe intentions rather than changes.

The framework most buyers recognise, and the one that structures this article, is the quality management standard published by ISO. Alignment with it is not a certificate requirement; it is a way of knowing the system is complete.

The Documentation Hierarchy: Specification, Procedure, Record

An assurance system has three document layers and confusing them is the most common structural weakness. Each layer answers a different question and each has a different revision rhythm.

The specification states what the product is. The tech pack, the material specification, the dimensional drawing with tolerances, the colour standard and the approved sample together define the requirement. The specification changes when the product changes.

The procedure states how the work is done. The assembly method, the inspection method, the test method and the packing method describe the how, and they change when the method improves rather than when the product does.

The record states what happened. Inspection results, test reports, training records, approvals and corrective actions are records, and they are never revised, only added to.

The hierarchy matters because a change at one level requires a check at the others. A specification change requires a check that the procedure still produces it and that the record format captures the new characteristic.

Revision control is the discipline that makes documents trustworthy. Every document needs a revision identifier and a date, and the current revision has to be the one in use on the line. Old revisions physically present in a work area are the most common source of a wrong-version error.

Accessibility is the second discipline. A specification that only exists in a buyer's email is not controlled, and a procedure that only exists in a supervisor's memory is not a procedure.

Retention is the third. Records are retained, procedures are retained as history, and superseded specifications are retained for traceability. Discarding a superseded specification destroys the ability to answer a question about a past shipment.

The practical test is simple: ask for the current revision of one specification, one procedure and one record, and see how quickly all three arrive with consistent references. The speed of that response is a reliable indicator of system health.

First Article Approval and the Golden Sample

First article approval is the gate between sampling and bulk, and it is the single most valuable assurance process in a soft goods programme. It establishes the physical reference against which everything that follows is judged.

The first article is a unit produced using the bulk tooling, the bulk materials and the bulk method, not a prototype made by a skilled sample maker. That distinction is the entire point: the first article predicts bulk output, a prototype does not.

The approval covers four things: dimensional conformance against the specification with tolerances, material conformance against the approved components, functional performance of every zip, buckle, adjuster and closure, and appearance against the approved sample.

Dimensional measurement should be recorded rather than observed. A measurement sheet with the specified value, the tolerance and the measured value for each critical dimension is the artefact, and it is the document that resolves a later disagreement about fit.

Functional testing of the first article should be more thorough than any subsequent inspection. Every closure cycled, every strap loaded, every attachment point pulled, and the results recorded. A first article that has not been functionally exercised has not been approved.

The golden sample is the physical outcome. A unit from the approved first article run, signed, dated and stored, is the reference for production, for inspection and for any dispute. It should be held by both parties.

Storage conditions matter more than expected. A fabric shade changes with light exposure and a foam compresses with stacking, so a golden sample stored on a sunny shelf or under weight becomes an unreliable reference within a season.

Re-approval triggers should be stated. A material change, a component change, a tooling change, a method change or a specification change each require a new first article. Without stated triggers, the golden sample quietly becomes out of date while still being used as the reference.

Timing sits inside the standard windows: samples in 6-10 working days, bulk production in 35-50 days after approval at MOQ 500 pieces per colourway. First article approval is the event that starts the bulk clock, so its own delay is the most expensive delay in the programme.

Change Control and the Engineering Change Notice

Uncontrolled change is the largest single source of quality failure in a repeat programme. Everything else in an assurance system can be working perfectly and a single unrecorded component substitution will produce a season of defects.

The process is straightforward: a proposed change is documented, assessed for its effect on conformance, approved by the parties who own the requirement, implemented with a stated effective point, and verified. Five steps, one document.

The effective point is the detail that prevents most damage. A change should take effect at a defined lot or date, and goods produced before that point remain to the old specification. Without a defined effective point, a mixed lot appears.

Assessment should cover more than the changed item. A new webbing affects the seam that holds it; a different foam affects the assembled dimension; a new zip affects the panel geometry. Change assessment is a system question, not a component question.

Approval authority should be explicit. A change affecting safety, compliance, function or appearance should require the buyer's written approval, because those are the characteristics the buyer is contractually responsible for.

Verification closes the process. After implementation, the first lots under the change should be checked against the revised specification, and the result recorded. A change without verification is a hypothesis.

Documentation of the change should reach the inspection checklist. A changed characteristic needs a revised acceptance criterion, and an inspection working to the old checklist will not detect a failure introduced by the change.

The commercial protection this offers a buyer is significant. Where a defect appears in season two that was not present in season one, the change control record is the first place to look, and its absence is itself the answer.

The discipline applies to buyer-initiated changes too. A buyer requesting a cost reduction should expect the same assessment, because a cheaper component is a change with consequences.

CAPA: Corrective and Preventive Action

Corrective and preventive action is the process that converts a failure into a permanent improvement, and it is the assurance process most often performed badly. The usual failure is that a corrective action is written but never verified.

Corrective action addresses a nonconformity that has occurred. Preventive action addresses a risk that has been identified before it occurs. Both use the same structure and both need the same closing discipline.

Root cause is the step that is skipped. Treating a symptom, replacing the defective units, retraining an operator, is a correction, not a corrective action. A root cause asks why the process allowed the defect, and it usually points at the specification, the method or the material.

A simple technique is to ask why five times. A strap failed because the seam pulled; the seam pulled because the allowance was insufficient; the allowance was insufficient because the specification did not state it; the specification did not state it because nobody owned that field. The fourth answer is the root cause.

The action must be specific and owned. An action with a named owner and a date is a commitment; an action described as increased awareness is not an action at all.

Verification is the closing step and the one usually missing. After the action, the specific defect should be measured on subsequent lots to confirm it has reduced, and the measurement should be recorded. Without verification, a corrective action is an assertion.

Effectiveness review is the final step. If the defect rate has not changed, the root cause was wrong and the process starts again. A programme with many closed corrective actions and a flat defect rate has a root cause problem, not an action problem.

Preventive action uses the same structure on a risk before it materialises. A new operator, a new material, a tightened schedule and a new style are all triggers for a preventive review, and a short meeting before each is cheaper than a corrective action after.

External verification is available where a buyer wants independence on a persistent problem, and testing and inspection providers such as SGS can assess a corrective action and report on its effectiveness.

Internal Audit and Management Review

An assurance system that is never audited drifts. Internal audit is the process that checks whether the documented system is the system actually being followed, and it is the cheapest way to find drift before a customer or a retailer does.

Internal audit checks three things: whether the documents are current, whether the records show the work being done as documented, and whether the work produces the intended result. Three questions, applied to each process in rotation.

Independence within the audit matters. An auditor auditing their own work will find little, and even a small organisation should rotate auditors between areas rather than assign each person to their own process.

Frequency should be scheduled rather than reactive. A full cycle over a year, with higher-risk processes audited more often, is the normal pattern, and an audit triggered only by a failure is not an audit programme.

Findings need the same treatment as any other nonconformity: a root cause, an action, an owner, a date and a verification. An audit finding without a corrective action is a record of a known problem that nobody fixed.

Management review is the second half and the one most often omitted. A periodic review by the people with authority to allocate resources, considering audit results, defect trends, corrective action status and customer feedback, is what turns data into decisions.

Input to management review should include the buyer's own data. Complaint rates, return rates and inspection outcomes from the customer side are the most valuable input a supplier receives, and a review that excludes them is incomplete.

Output should be decisions with owners: a specification change, a training programme, an equipment purchase, a supplier change. A review that produces minutes but no decisions is a meeting.

Buyers should ask when the last internal audit and management review took place. The dates, and the actions that came out of them, tell a buyer more about a quality system than any certificate.

Supplier and Subcontractor Assurance

A pet bag is an assembly of components made by other people, and the assurance system has to extend beyond the assembly operation to be worth anything. Fabric, webbing, foam, hardware, labels and packaging all come from upstream suppliers, each with its own process variation.

Qualification is the first step and it is a one-off per supplier per component type. A new component source should be qualified on samples before being used in bulk, with the same checks applied as for a first article.

Ongoing evaluation is the second. A simple scorecard covering incoming conformance, delivery reliability and responsiveness to problems, reviewed periodically, is sufficient and is far better than no measurement.

Certificates and test reports should be requested for the characteristics that matter. A material certificate, a substance test report or a cycle test result is the evidence behind a specification claim, and the claim should not be made without it.

Subcontracted processes need the same treatment. Where printing, welding, coating or a specialised assembly step is subcontracted, the subcontractor's process is part of the product's process and should be covered by the assurance system.

Dual sourcing has an assurance implication that buyers often miss. Two approved sources for the same component are only equivalent if both are qualified and both are held to the same specification, and in practice one is usually tighter than the other.

Change notification from upstream should be contractually required. A component supplier changing a formulation or a process without notice propagates a change into the finished product with no assessment, and that is the uncontrolled change described earlier.

Traceability closes the loop: knowing which upstream delivery went into which production lot is what makes any upstream problem containable.

Independent verification of upstream claims is available and worth using for the components that carry safety or compliance risk. Bodies such as ASTM International publish the test methods, and testing providers can verify conformance to them where a claim matters commercially.

Risk-Based Thinking: Where Assurance Effort Should Concentrate

An assurance system applied uniformly across every characteristic is expensive and less effective than one concentrated where the risk actually sits. Risk-based thinking is the discipline of allocating assurance effort in proportion to consequence, and it is the difference between a system that works and one that is merely documented.

The first question for any characteristic is severity of failure. A strap that fails under load, a tether that releases and a ventilation panel that is blocked all carry consequences far beyond a cosmetic shortfall, and each warrants a heavier control than its share of defect volume would suggest.

The second question is detectability. A defect that is invisible in the finished unit, an incompletely secured strap end, a coating that will delaminate later, or a shade difference only visible in assembly, cannot be managed by final inspection and needs control at an earlier gate.

The third question is stability. A characteristic that drifts with material batch, operator, machine setting or season needs frequent monitoring; one that is stable across lots can be checked far less often without added risk.

The fourth question is the cost of the control relative to the cost of the failure. A pull test on every restraint attachment is cheap against the consequence of a failure and expensive against the volume, which is why a stated frequency rather than a full check is usually the right answer.

Applying the four questions produces a ranked list, and the top of that list is where the specification detail, the SOP coverage, the inspection frequency and the corrective action priority should all concentrate.

The method also identifies where effort can be reduced honestly. A stable characteristic with low severity and easy detectability does not need heavy control, and releasing that effort is what funds the concentration at the top of the list.

Review of the ranking should be periodic and evidence-driven. A characteristic that has generated no findings for several seasons can be de-escalated, and one that has produced a claim should be escalated immediately.

The framework for this approach is embedded in the modern quality management standard published by ISO, and a buyer asking a supplier how it allocates assurance effort will learn more from the answer than from any certificate.

Certifications, Audits and What a Buyer Should Request

Certificates are the currency of supplier assessment and they are frequently over-read. A certificate states that a system was found to conform on a given date by a given body; it does not state that every order will be good.

ISO 9001 is the general quality management certification and the baseline expectation for a wholesale programme. It confirms that documented processes exist and are maintained, which is useful and insufficient on its own.

Social compliance audits, such as the BSCI framework, address labour and ethical standards rather than product quality, and most retailers require them separately. They should not be confused with a product quality audit.

Product and material certifications address the article itself: substance compliance, restricted substance testing, and any market-specific conformity requirement. These are the certificates that determine whether goods clear a border or a retailer's onboarding.

Third-party inspection reports sit alongside certificates as the current evidence. A certificate is a statement about a system in the past; an inspection report is a statement about a lot in the present, and a buyer should have both.

Audit reports should be requested in full rather than as a certificate alone. A report showing findings, corrective actions and their closure tells a buyer far more than a pass statement, and a supplier willing to share it is signalling confidence.

What a buyer should request is therefore a short, specific list: the current certificate with scope and validity, the most recent internal audit summary, the corrective action log with verification status, and the per-lot inspection summary. Four documents.

The request itself is a test. A supplier that produces those four documents within a day is demonstrating the system whether or not the documents are perfect; one that takes three weeks is demonstrating something else.

Where a programme serves a regulated market, the product safety expectations should be reflected in the checklist, and the framework published by the CPSC is a reference point for US-facing goods.

Finally, align the request to the order terms. Asking for the documents as a condition of release, rather than as a favour afterwards, is what turns an assurance system from a claim into a commercial commitment.