Quality control · inspection points · defect classes · reject rates · remedies

Quality Control and Rejects in Custom Merch

Most quality problems in made-to-order merchandise are not caused by an incapable supplier. They are caused by an inspection that happened at the wrong moment, or not at all. A colour that was never checked against the approved reference cannot be disputed once the run has shipped. A count verified at the carton but never against the packing list cannot be reconciled. A defect classification that was never agreed leaves both sides arguing about what counts as a defect while the goods sit in storage. This guide sets out what can actually be checked at the sample, pre-production and delivery stages, how to define defect classes so the argument does not happen later, how reject rates and remedies are agreed in advance rather than after the fact, and what a buyer has to write down to have any chance of a remedy at all.

Request a supplier pack Read all guides

Three inspection points, and the failure each one catches

The sample stage settles direction rather than the item. It is where colour placement, decoration position and proportion, the tone of a photograph against a print, and the feel of a chosen closure are decided. A sample cannot confirm what a production run will do, because it is made under different conditions — a different batch of blanks, a single unit rather than a run, frequently a different operator. What it does is move the decision earlier, while changing it is still cheap. The failure it catches is the most expensive one in the category: a design that was never right. Money spent on a sample before production is the cheapest money available anywhere in this process, and money refused on a sample is not lost at all.

The pre-production stage is the one most often skipped, because it sits in a window that is inconvenient for everyone involved. Its purpose is to establish what the run itself is doing rather than what the specification says it should do: the production units photographed or sampled against the approved proof, the colour reference actually in use, the grade of blank actually drawn from stock. The failure it catches is silent drift — a grade substituted, a reference shifted, a placement moved slightly for a reason nobody wrote down. Each of those is invisible on the invoice and obvious on the goods, which is why it can only be caught here and nowhere later. This is also the last point at which a run can be stopped, which makes it the only stage where a finding still changes the outcome rather than merely documenting it.

The delivery stage is inspection at receipt against what was written: count against the packing list, decoration against the approved proof, substrate against the stated grade, colour against the named reference, packaging as specified, and units per size or recipient exactly as delivered. What it cannot do is catch anything that needed catching earlier, because by this point the money is spent and the run is finished. Its real purpose is to preserve the remedy rather than to prevent the problem. A discrepancy found at receiving can still be raised against the order, with the specification, the proof and the goods all in one place; the same discrepancy found at month end can only be raised as a credit request against an invoice, which is a slower and weaker position.

Sample stage: Direction: colour, placement, proportion, feel — confirms the design, not the production run
Pre-production stage: The run about to be produced: grade, colour reference, placement, packaging — the last point at which it can be stopped
Delivery stage: Count, decoration, substrate, colour, packaging against the written specification — preserves the remedy rather than preventing the problem
Evidence retained: Approved proof, written specification, colour reference, inspection note, photographs, packing list, delivery note
Tolerance: Stated per parameter before production; a tolerance nobody set in advance is a dispute waiting for a reason

Four defect classes, and what each one is actually worth

Most reject-rate arguments are really classification arguments in disguise: the parties already agree that something is wrong and are arguing about what to call it. Fixing the classes before production removes that argument and leaves only the measurements. Four classes cover nearly all of what occurs in made-to-order merchandise, and each carries a different commercial weight.

Critical: unusable or unsafe, and the tolerance is zero

The class defined by consequence rather than by appearance: an item that cannot do the job it was commissioned for, or that presents a risk to the person handling or using it. Sharp edges on a hard good, a closure that fails under load on a bag intended to carry something, a food-contact item that fails a required check, an item intended for children produced with a material never approved for that use. The commercial rule for this class is a stop rather than a percentage: a run containing a critical defect is not shipped, and it is not resolved by sampling around it, because a critical defect is a statement about the process that produced it rather than about the individual unit. Critical defects are rare, and they are rare precisely because everyone understands the tolerance is zero. What damages a buyer more than a critical defect is discovering one after the goods have been distributed.

Major: the recipient would notice it differs from what was approved

This is the class that carries almost all of the commercial weight, and it is where most reject-rate clauses are rightly concentrated. Colour outside the agreed tolerance of the named reference, a substrate below the stated grade, decoration misplaced or absent across a defined share of units, print adhesion that does not survive handling, a seam or closure that fails, a size split delivered as something other than what was ordered. The common thread is not severity but visibility: someone outside the project would see that these units are not the units that were approved. Major defects are also the class most reliably measurable, because each one can be counted on a sample against a written criterion, which is exactly why they should carry a stated threshold and a stated disposition rather than an assurance that the goods will be correct.

Minor: visible, but not functionally significant

Light surface marks, print registration variance within reason, a slightly loose thread, an edge on a print area, a small dimensional difference that falls inside a stated tolerance. The judgement here is contextual and must be recorded as contextual, because a mark on an item handed to a client in a meeting room is not the same mark on an item issued to every employee on an open floor. A class defined as minor without stating where it applies becomes a class that is argued about at receipt, usually by whoever received the delivery. Two things make this class workable in practice. Define it against the tolerance already written into the specification, so there is nothing to argue about. And record minor defects by programme rather than by order, because a repeated observation across three deliveries is a process signal and a single observation on one delivery is not.

Packaging and handling defects

The class most often excluded from reject-rate arithmetic, which is a mistake on hard goods. Crushed cartons, units broken in transit, individual presentation absent from a kit that specified individual presentation, mislabelled cartons, items packed in quantities that make distribution to recipients impossible. Breakage in transit can exceed the defect rate of the print itself by a wide margin, and it is the class the buyer effectively pays for twice — once for the damaged units and again for the replacement units ordered at short notice into the same programme. Handling defects are also governed by a different part of the commercial arrangement from print defects, since the packaging specification and the transit protection are set separately from the decoration. Name the class, specify the transit packaging, and record the count at receiving; leaving it unnamed means the most expensive failures are the ones nobody agreed to measure.

Agreeing a reject rate before the run, not after

A reject rate is a number with a method attached, and without the method it is a number with an argument attached. The pieces that have to be present are unglamorous and all of them change the result. What is the unit of count: every unit produced, or every unit delivered, which differ whenever units are rejected before dispatch. How many units are actually inspected: full count on a small run, sampling on a large one, with the sampling method stated rather than implied. Who inspects, and whether the supplier may attend, since an inspection conducted alone by one interested party will be questioned later if anything goes wrong. Against what written standard the units are judged, which is the specification rather than a general expectation of good quality. The threshold itself, expressed as a percentage against a named defect class rather than against the order as a whole. And the disposition, which is where the clause either works or does not.

Disposition is the part that gets skipped, and it is the part that decides the outcome. Four dispositions exist in practice and a clause should name which one applies rather than inviting negotiation at the moment of maximum pressure. Rework by the supplier at their own cost, which suits units with a minor or major defect that can be corrected without remaking them. Replacement units supplied within the same run, which is the practical answer when rework is impossible and the programme has a fixed distribution date. A credit against the affected units, which is the cleanest outcome where the buyer has already decided how to proceed without them. Or acceptance with a recorded concession, where the affected quantity is small enough that reworking or replacing it would cost more than accepting it, and where accepting it is a decision rather than an oversight. Each implies something different commercially. Leaving disposition open means the two parties settle it while the goods are sitting in a warehouse.

Two refinements make the clause easier to apply than a single flat number. Set the threshold per defect class, so a run carrying a small proportion of minor marks is not treated as a failed run, and so a run containing any critical defect is not rescued by an average that sits comfortably inside tolerance. And write the measurement into the clause rather than assuming it — a percentage means nothing without stating what was counted, how many units were inspected and against which standard. Suppliers accept this more readily than buyers expect, because a method nobody has written down is a method either side can reinterpret once the outcome is known.

What the buyer documents, and why it decides the outcome

An observation only becomes a claim once it has become a record. Four documents carry almost all of the weight, and none of them is expensive or difficult to produce. The approved proof, which defines what everyone agreed the item should look like. The written specification, which defines what everyone agreed would be delivered — grade, decoration method, colour reference, placement, quantity, packaging. The colour reference actually used, recorded at the moment of approval rather than reconstructed afterwards, because a reference named three weeks after the run is a reference chosen by whoever remembered it. And the inspection note, which is the document that turns the first three into something actionable.

The inspection note is the one most often missing and the one that decides the outcome. Four lines are sufficient: what was inspected and how many units, against which standard, what was found and in which defect class, and what was accepted. Photographs of the affected units belong next to the approved proof, because a print variation is difficult to describe in words and trivial to show. Note the part buyers find counter-intuitive: an inspection note recording a pass is as valuable as one recording a failure. It establishes that the goods were examined at a defined moment against a defined standard by a named person, which is exactly what an acceptance clause requires before it can operate, and it is the document that answers any later question about whether a defect was raised at the time or manufactured afterwards.

Keep the file with the order rather than with the person who inspected. If the record lives in an inbox and that person later leaves, the buyer’s position on a disputed delivery rests on recollection, which is the weakest evidence available and the one a supplier is entitled to decline to accept. The standard applied to the record should not vary with the outcome: the value of the file is that somebody who was not present can read it, reconstruct what was agreed and what arrived, and reach a defensible conclusion without asking a single follow-up question.

Questions buyers ask about quality and rejects

When should we inspect a merchandise delivery, and how many units do we check?

At receipt, before the goods are put into storage or distributed, and against the written specification rather than against a general impression. On quantity, full count where the run is small enough that it is affordable, and a stated sampling method where it is not — with the method written into the specification or the agreement so that both sides know what was checked. What matters more than the proportion inspected is that the inspection is recorded. A note stating how many units were examined, what was examined, what was found and what was accepted closes the question; an inspection performed and not written down leaves the buyer describing a conversation at a point where the supplier reasonably asks to see the record.

How do we agree a reject rate with a supplier without it becoming a negotiation?

Agree it before production, against a written method, and write the disposition down at the same time. The three elements that remove almost all of the friction are the unit of count, the sampling method, and the remedy. Say whether the threshold applies across all units produced or all units delivered, say how many units will be inspected and how, and say what happens to the affected units — rework, replacement, credit or recorded concession. Once those three are written, a defect count stops being an event and becomes an input to a rule. What turns the discussion into a negotiation is leaving disposition open, because then every delivery raises a fresh question about what should happen rather than applying an answer both sides already agreed to.

What counts as a major defect rather than a minor one?

Visibility to a recipient is the working test: if someone outside the project would notice that the item differs from what was approved, it is major. A colour outside the agreed tolerance of the named reference, a substrate below the stated grade, decoration misplaced across a share of units, print adhesion that fails under handling, a closure or seam that gives way — all major. A light surface mark, registration variance inside the stated tolerance, or a small dimensional difference within an agreed band is minor. The distinction only works if the tolerance was written before production, because without a stated band there is no objective point at which visibility becomes a defect. And it is worth deciding per programme where minor defects are acceptable, since the same mark means different things on an item handed to a client and on an item issued to every member of staff.

Is Corp-Merch affiliated with, endorsed by, or an official supplier to the brands, agencies, or institutions referenced on this page?

No. Corp-Merch is an independent B2B merchandise supplier. Any mention of a manufacturer, a brand, a standards body, a carrier or an institution is informational and implies no partnership, sponsorship, endorsement or official-supplier relationship. Nothing on this page is legal, tax or financial advice, and any threshold, tolerance or remedy described here reflects typical practice for this category rather than a fixed commercial offer.

This page is written for procurement, quality and operations teams controlling custom corporate merchandise produced in Armenia. Corp-Merch is an independent B2B merchandise supplier. The inspection points, defect classes, thresholds and dispositions described here are a reasoned framework for managing quality in this category; they are not legal, tax or financial advice, and an organisation should adapt them to its own internal controls and to the requirements of the goods in question. Any delivery term, payment term or lead time referenced describes typical practice in this category rather than a fixed quotation. Each order is priced individually against a written specification, an approved proof and a signed agreement.