How many advertised discounts are real?

A crossed-out price is a claim, and a claim can be checked. We read 475 shops twice a day, so when one of them announces a markdown we can ask the only question that matters: is anyone else charging less right now? That splits every announced discount into three, and the split is not close.

• Real — the shop marks it down, and no competitor we read undercuts it. 2,209 today, 35.5% of those we can judge. These are worth acting on, and they are the minority.
• Inflated — the discount is real but smaller than announced: the struck-through price is above what the market actually charges, so the saving is measured from a number nobody asks. 1,648 today.
• Beaten — the shop advertises a sale while another shop's plain, unadvertised price is lower. 2,366 today.

Put the last two together and you get 64.5% of every advertised discount in the corpus. That is not a claim about intent. A merchant can set a reference price in good faith and simply be more expensive than a rival he does not watch. But the reader who sees “30% off” and assumes it is a good price is wrong more often than right, and no single shop's page can tell him so.

The gap between the two columns below is the whole subject of this page. The left column is what shops say. The right is what we can defend when the inflated ones are measured against the cheapest competitor actually charging that day.

The right column is computed only on the 1,648 inflated discounts, because those are the ones where a real alternative price exists to measure against. A “beaten” discount has no meaningful percentage: the honest statement there is not “it is worth less”, it is “somebody else is cheaper, here is who”.

The widest gaps we can currently document, in the shop's own currency. Each row is one shop advertising a markdown while another shop, the same day, quietly charges the figure in the last column.

The full live list is at fake sales, rebuilt twice a day.

• A discount under 3% and under $1 is not a discount. A shop that crosses out five cents is not claiming anything, and flagging it would bury the cases that matter. That threshold is why the count here is smaller than the number of struck-through prices in the catalogue.
• We compare independent sources, not listings. Two storefronts running the same catalogue are one source; counting them twice would let a merchant confirm his own price. The comparison needs a genuinely separate shop.
• A pair whose prices differ by more than a factor of four is thrown out. Nobody sells the same object at $22 and at $3,250 — that is a matching error, not a lie, and publishing it as a lie would be our error rather than the shop's.
• We never say a shop is dishonest. We say what two shops charge on the same day, and link to both so the reader checks without trusting us.
• 35.5% of announced discounts hold up, and we say so as plainly as we say the rest. A page that only found villains would be a page looking for them.

Is a bigger advertised discount more likely to be fake?

Not reliably, and that is the uncomfortable part. The median announced discount across all 6,223 judgeable claims is 30%, and claims at the top of the range — 50% and above — are neither notably more nor notably less likely to be beaten by a competitor. A large percentage tells you which reference price the shop chose, not what you will pay.

What is the difference between an inflated discount and a beaten one?

An inflated discount is measured from a number nobody charges: the struck-through price sits above what the market actually asks, so the saving is real but smaller than advertised — 1,648 today. A beaten discount is one where a different shop's ordinary, unadvertised price is lower than the sale price — 2,366 today. The first overstates a saving; the second offers one that is not the best available.

Does this mean shops are lying about their prices?

No, and we do not say so. A merchant can set a reference price in good faith and still be more expensive than a rival he does not track. What we publish is what two shops charge on the same day, with a link to both, so a reader checks instead of trusting. 35.5% of announced discounts hold up under exactly the same test.

How can a price tracker know what a discount is really worth?

Because the question is comparative, not historical. We do not need to know what a product ought to cost — only what somebody else is charging for it right now, and we read 475 shops twice a day to find out. No single shop can run this test on itself at any price.

Why are there fewer judgeable discounts than crossed-out prices?

Two filters. A markdown under 3% and under $1 is not treated as a claim at all — a shop crossing out five cents is not asserting anything. And a discount can only be judged where a second independent shop carries the same product, which is true of a minority of any catalogue.