What share of memory-card returns is actually preventable?

In practice, the large majority. Counterfeit/fake-capacity returns are removed by sourcing from a verifiable factory rather than pooled marketplace inventory and testing the batch. Wrong-card returns (a consumer card sold for a dashcam, or an A2-but-low-endurance card for continuous recording) are removed by matching the spec to the use at the counter. What's left — true random failures on genuine, correctly-specced, endurance-grade stock — is a small remainder a clear warranty covers.

How do I stop selling fakes without testing every single card?

Move the testing upstream. Buy from a source that 100% capacity-tests each batch at the factory and ships the report, so the guarantee arrives with the goods. Then spot-check incoming lots yourself with H2testw or F3 on a sample, not the whole shipment. The point is provenance plus a sample check — not testing every unit at your counter.

A customer's card "keeps dying" — is it defective?

Usually it is the wrong card for the workload, not a defect. A dashcam or security camera writes tens of terabytes a year; a consumer card with no endurance rating wears out in months under that load. Before processing it as defective, check what device it's in. If it's continuous recording, the answer is an endurance-grade card with a stated TBW — and that's an upsell, not a refund.

Why memory cards come back — and how a reseller cuts the returns

If you sell memory cards, you already know returns are the quiet tax on the business. A return isn't just the unit you swap or refund — it's the customer who now wonders whether to come back, and the time your counter spends instead of selling. The good news is that returns aren't random. Pull the thread on a stack of them and almost all trace to three causes, and each one is prevented before it ever reaches your counter.

A return costs you twice

Count the real cost. You lose the unit (or eat the refund and the shipping), and you spend staff time on the swap. But the bigger cost is the relationship: a customer who got a fake or a card that died once tends to buy the next one — and the next camera, and the next bulk order — somewhere else. Cutting returns isn't a service nicety; it's margin and retention. And the cheapest place to cut them is upstream, not at the counter.

Cause 1 — counterfeit / fake capacity

The single most damaging return, because the customer loses data, not just money. A counterfeit card reports a fake size and silently overwrites once you pass the real capacity, so it tests "fine" at first and fails later — often after your return window, which is worse for trust even when it's not your fault.

This is a sourcing problem, not a pricing one. It enters through pooled, commingled marketplace inventory where anyone's returned fake can land in a genuine listing. Prevent it by:

Buying from a known factory source, not a marketplace lot of mixed provenance.
Sourcing stock that is 100% capacity-tested at the factory with a report, so the guarantee arrives with the goods.
Spot-checking incoming lots yourself with H2testw or F3 on a sample — see how to test for fake capacity.

Cause 2 — the wrong card for the job

The customer's card "keeps dying," so they bring it back as defective — but it was never the right card. The classic case is a consumer card sold for a dashcam or security camera, where continuous recording writes tens of terabytes a year and a card with no endurance rating wears out in months. (The write-load math is in choosing a card for dashcams and security cameras.)

This is a point-of-sale problem, and fixing it is an upsell, not a refund:

Ask what device the card goes in before you sell it.
Recorder? Sell an endurance grade with a stated TBW or rated hours.
Video in a camera/drone? Match the V30/U3 class to the bitrate; don't oversell A2, which does nothing for sequential recording (speed classes decoded).
Phone, handheld, single-board computer? That's where A2 earns its price.

Match the spec to the use and the "defective" return becomes a correct — and usually higher-value — sale.

Cause 3 — genuinely low endurance or DOA stock

What's left after fakes and mismatches is real quality: cards that are genuine and correctly specced but still fail early because the underlying flash and firmware are bottom-tier, or arrive dead on arrival. This is again a sourcing decision:

Buy endurance-graded product with a published TBW, not "endurance" as a bare label with no number.
Source from a factory that does outgoing test and stands behind a concrete warranty term (years), with a clear, fast replacement path — ideally one you can run locally rather than shipping every claim back overseas.

A pre-sale checklist

A 20-second exchange at the counter removes most of Causes 1 and 2:

What device? (recorder / camera / phone / handheld / PC)
Continuous recording? → endurance grade + stated TBW, V30, wide temperature.
Running apps or an OS? → A1/A2.
Capacity sized to the need, with the real usable figure in mind (a "128 GB" card is ~119 GiB).
Genuine? Sold from tested, factory-sourced stock — with a report you can show.

Where sourcing does the heavy lifting

Two of the three causes are sourcing decisions, and the third is a counter habit. Get the supply chain right — verifiable, endurance-graded, factory-direct, with outgoing capacity tests and a warranty you can honour locally — and the return pile shrinks to the small, genuine remainder that any clear warranty absorbs.

That's the side of it we're built for: factory-direct stock, every batch 100% capacity-tested with the report shipped alongside, endurance grades with the TBW stated up front, and samples to verify before you commit. Tell us what you sell and to whom, and we'll spec a line that comes back less. For the sourcing due-diligence side, see how to vet a flash-memory supplier.