How to recognise a fake discount: five tests that take a minute

"−70 %" from a price nobody paid is not a discount. Five one-minute tests — sanity, history, siblings, countdown, anchor — plus the rules NextToolHub applies before a deal is shown.

A fake discount is not usually a lie about the current price; the current price is real. The lie is the reference price — the crossed-out number the discount is measured against. Sellers know that '−70 %' converts better than 'USD 12', so the reference price drifts upward until the percentage looks dramatic. The five tests below expose that in about a minute, without any tool.

Test 1 — the sanity check

Ask one question: would anyone have paid the crossed-out price for this? A generic USB cable 'was USD 39.90, now USD 3.99' fails instantly — no USB cable of that kind has sold for USD 39.90 in years. If the reference price would make the item the most expensive of its category, it was never a price.

In our own pipeline the mechanical version of this test is the ratio limit: a list price six times the current price is treated as a feed error and the discount is discarded, not shown.

Test 2 — the price history

A real discount is a drop from a price the item actually sat at. Marketplaces do not show history, but you can approximate it: the item's price on a comparison site a week ago, cached search results, or simply your own memory of the category. On NextToolHub the deals page only shows discounts whose feed data is younger than seven days and records when the list price was first seen — an item whose 'original' price appeared for the first time today, together with its discount, is a red flag.

If the 'sale' has been running for months, it is not a sale; it is the price with theatre.

Test 3 — the sibling test

Open two or three other listings for the same or a near-identical product. If their normal prices cluster around the 'discounted' price of the first one, the discount is the market price wearing a costume. Genuine discounts stand out against siblings; fake ones match them.

This is the fastest test on marketplaces, because near-identical listings from different sellers are always a search away.

• Same product, three sellers, three 'normal' prices — the median is the real price.
• A 'discount' landing exactly on that median is not a discount.
• Wildly different reference prices across siblings mean the reference is invented.

Test 4 — the countdown test

Timers that reset when the page reloads, 'only 3 left' that never becomes 2, and 'sale ends tonight' every night are pressure devices, not information. They do not prove the discount is fake, but they tell you the seller is optimising for panic rather than for a fair comparison — and fabricated urgency correlates strongly with fabricated reference prices.

Reload the page once. If the timer restarts, treat every other claim on the page with the same scepticism.

Test 5 — the anchor test

Bundles and multi-packs often show a reference price that is the sum of buying each part separately at full price — a number nobody would pay. Similarly 'MSRP' on a marketplace item with no brand is meaningless: there is no manufacturer setting a suggested price. Ask what the reference price is anchored to; if the answer is 'nothing verifiable', ignore the percentage and judge the current price on its own.

This is why NextToolHub labels the source of every list price ('store feed list price') and shows when it was first seen, instead of just printing a percentage.

Method

These tests are the ones our deal policy code approximates automatically (recomputed discount, ratio limits, freshness window, manual approval above 50 %) and the ones we apply by hand when reviewing that approval queue. They are written for marketplaces such as AliExpress and Alibaba.com, where the reference price is set by the individual seller, but they work anywhere a crossed-out price appears.

Verdict

Judge the current price on its own merits first; only then ask whether the reference price is real.

The sibling test settles most cases in under a minute: three sellers, three normal prices, one median.

A discount that survives the sanity check, has a real history and a verifiable anchor is a discount; the rest is presentation.

Questions people ask

Does NextToolHub show the seller's own discount percentage?

No. We recompute it from the feed's list price and current price, discard ratios above 6×, and hold discounts above 50 % for manual review before they can appear on the deals page.

Why do I see fewer deals on NextToolHub than on the store's own sale page?

Because most of the store's sale page fails the tests above. We would rather show 40 discounts you can trust than 400 you cannot.

Can I report a fake discount?

Yes — use 'Report a wrong price or product' on the product page. Reports go to a queue we review, and the listing's list price is re-checked.