Last updated: 28 July 2026. This page is sourcing and quality guidance from a trading company — it is not legal advice, and it is not a substitute for your own testing, certification or compliance programme.
Quality fade is drift, not failure. Each shipment sits a little further from the specification than the one before it — by an amount small enough to pass on its own, and large enough to matter by the sixth order. Below is how the substitution actually creeps in, why it catches experienced buyers rather than beginners, and the controls that stop it.
How it actually happens
Where the substitution creeps in, order by order.
The reason quality fade is so hard to argue about with a supplier is that no single step in it looks unreasonable. This is the shape it usually takes.
- Order one. You approve a first-article sample. The factory is on its best behaviour, an engineer is watching the line, and the material is exactly what was quoted. This unit is the only honest reference point you will ever have — which is why what happens to it matters more than anything else on this page.
- Order two. A pilot run against that sample. Clean. Trust is established, and your inspection intensity begins to fall, because caution has now been unrewarded twice.
- Order three. The factory's own input costs move. A resin, a steel grade, a coating, a fastener — one of them now costs more than it did when they quoted you. Your price did not move. Something has to.
- Order four. A component you never see gets changed: an inner layer, an adhesive, a plating thickness, the grade of a screw. It is chosen precisely because it survives the checks you actually run. It looks identical, weighs nearly the same, and fails later — at your customer's house rather than on an inspection table.
- Order five. Something arrives slightly out of spec. Wall thickness a hair under, fabric a little lighter, finish marginally rougher. It is minor and the stock is needed, so it ships. Nobody tells the factory it was an exception.
- Order six. Your new baseline is order five. The next batch starts from there and drifts the same small distance again. Put order one and order six side by side and you are holding two different products — bought under one specification, at one price, from a supplier you vetted correctly.
Why it survives. Every step is small enough to defend on its own, and every shipment gets compared against the one before it rather than against the original. A failure gets caught. A drift doesn't.
Why it catches good buyers
Factories get better at production. Specification tends to go the other way.
The reasonable expectation is that a factory gets better at your product as it makes more of it. That is often true of process — cycle times, yield, handling, packing. It is far less reliably true of specification, because the two are driven by opposite incentives. Process improvement saves the factory money. Specification compliance costs it money.
The deviations are small enough to be invisible shipment to shipment and obvious only when order one and order six are on the same table. That is not a failure of attention. It is a failure of comparison basis, which is a much easier thing to fix.
And it can happen after years. Some of the worst cases sit inside relationships that ran cleanly for three or four years — long enough for the buyer to stop checking entirely. The drop, when it comes, arrives at whatever volume the relationship has grown to by then.
Where drift comes from
Two directions it comes from, and they need different answers.
In our experience the drift arrives from two directions. They look identical in the goods and are completely different to fix, so it is worth knowing which one you have before you open the conversation.
One — the factory quietly lowers its own bar.
Material substitution to protect margin as input costs rise. A cheaper component somewhere you can't see it. A sub-process handed to a workshop you have never audited. This is deliberate, incremental, and calibrated to sit just below your detection threshold. Nobody announces it, because announcing it would start the conversation they are trying to avoid.
Two — you allowed an exception and never said it was one.
An early shipment came in slightly off spec. The wall thickness a hair under, the fabric a little lighter than agreed. It was minor, so it shipped. That is not the mistake. The mistake is not telling the factory that it was an exception. Silence gets read as a revised tolerance, and the next batch starts from the revised one.
The second is the harder one to hear, because it means part of the drift was manufactured by the buyer's own reasonableness. It is also the one you can close this week, at no cost, with a single sentence in writing.
The stakes
Substitution goes where nobody is looking.
The components that get swapped are almost never the ones on the front of the product. They are the cheap, buried, unglamorous parts that no buyer writes an inspection line for — which is exactly why a swap there can run for months before anyone notices.
A documented case
Chinese carmakers Chery and Great Wall recalled roughly 23,000 vehicles in Australia after asbestos was found in brake linings — a low-value component buried inside a finished car, far below the level anyone was checking.
It is the clearest illustration of the mechanism there is. The car was inspected. The gasket-grade component inside it was not. The cost surfaced long after every one of those shipments had cleared customs, and it landed on the importer.
When to watch hardest
The weeks after Chinese New Year, and why operators plan around them.
Factories reopen at reduced capacity. Workers who went home for the holiday do not all come back — some take jobs closer to family, some leave the industry. Their replacements arrive knowing nothing about your product, your tolerances, or the exceptions your factory had learned to respect.
Accumulated knowledge about your specification lives in people far more than in documents. Every year, some of it walks out of the building.
We are not going to point you at a study that quantifies this, because we do not have one. What we can tell you is how experienced operators behave: they schedule an extra inspection in that window regardless of how reliable the supplier has been, and they treat the first post-holiday shipment as if it came from a new supplier's first run. It is one of the cheapest pieces of insurance in the sourcing calendar, and almost nobody buys it.
What actually works
Five controls. None of them expensive.
Compare against the original, not the last one.
Drift is invisible batch to batch and obvious across six. Keep the approved golden sample physically retained and inspect against it — not against the previous shipment, which has already moved.
Update the QC checklist before every order.
Reusing order one's checklist for order six means you are checking for last year's risks with last year's eyes. Every defect ever found should be added to it, permanently, so the checklist only ever gets harder to pass.
Re-test the things that are cheap to substitute.
Material grade, weight, thickness, coating, fill, component origin. Substitutions happen where they are invisible and where they save real money — which is a short and predictable list for any given product.
Say it out loud when you accept an exception.
In writing, once, every time: this is out of spec, we are accepting it on this occasion only, and the standard has not changed. This single habit closes one of the two directions entirely, and it costs nothing.
Make quality commercially consequential.
A small rebate for zero-defect shipments gives a floor manager a reason to care. A defined consequence for AQL failures above an agreed threshold — supplier covers re-inspection, discounts the next order — makes drift unprofitable. Factories respond to incentives far more reliably than to complaints, and both of these belong in the purchase order, not in an email.
One-off versus ongoing
A single inspection tells you about a single shipment.
That is worth having, and it is a sensible place to start. But quality fade is defined by change over time, so a one-off inspection can only ever tell you where you are — never which direction you are moving.
What detects it is continuity: the same inspector, checking against the same retained golden sample, with a checklist that accumulates every defect ever found, across every order. Continuity is the mechanism, not a packaging preference. It does not work as a series of unrelated visits.
That is the argument for a programme rather than a transaction, and we would rather state it plainly than imply it.
What we do about it
Six moves, in the order we would make them.
- Comparative inspection. Send us your first approved sample, or the retained golden sample, and we inspect a current production batch against it. You get a documented, photographed comparison of what has changed. This is the diagnostic, and it is the honest starting point.
- Checklist rebuild. We rebuild the QC checklist around what actually drifted on your product, not around a generic template.
- Retained golden sample. Held by us, in China, verified at every order. Not held by the factory, which defeats the purpose.
- Written exception protocol. Every out-of-spec acceptance documented and communicated to the factory as an exception. This alone closes one of the two directions the drift comes from.
- Scheduled high-risk inspections. After the Chinese New Year shutdown, and at any capacity change, ownership change or production-line move.
- Second source where it is warranted. A supplier who knows they are your only option has less reason to hold spec. We will find the alternative — and tell you when that is genuinely the root cause rather than a reflex.
Range
This isn't a category problem.
The mechanism does not change with the product. Only the specific thing that gets substituted does. Wherever a product has a material grade, a coating, a wall thickness, a fill weight or a component buried under an assembly, there is somewhere for a cheaper version to hide and a reason for it to end up there.
Tell us what you buy and we will tell you where substitution usually hides in it, and what it would take to check.
Engagement model
How this is priced.
- Comparative inspection US$268 per visit — one factory, one visit, photo-documented report within 24 hours.
- Checklist rebuild and golden-sample retention Included with an ongoing QC programme.
- Ongoing QC programme Monthly, priced on order frequency and supplier count.
- Post-holiday inspection slots Booked ahead of the Chinese New Year shutdown; the window is short.