We checked 24,690 Amazon deals over 15 weeks. Nearly 1 in 5 weren’t real.

A discount badge is the cheapest thing a retailer can print. It costs nothing to add “Save 30%” next to a price, and the “was” price it is measured against can be anything: a price the product held for a day, a price it never sold at, or a list price nobody pays. So we stopped asking whether a product was “on sale” and started asking a simpler question: is today’s price actually lower than what this product normally costs?

Since early July 2026 we have answered that question every week for a fresh sample of advertised Amazon deals. This article pulls the first 15 weekly reports together: 24,690 deal listings in total, from the week of July 6 to the week of October 19.

The headline: about 19% of “deals” were not deals

Across all 15 weeks, 4,625 of 24,690 advertised deals (18.7%) were not meaningfully below their usual price. That is close to one in five. These were not obscure listings or third-party oddities; they were products shown with a deal badge, a struck-through price and a percentage off.

The weekly share moved around more than we expected. The worst week was the week of August 3, when 27% of deals were not real. The best was the week of October 5, at 8%. Most weeks sat between 17% and 23%.

Each week links to its full report, which lists examples and lets you download the raw numbers.

Three things the trend tells you

1. Back-to-school was the worst stretch

The three highest weeks in the sample came back to back: the weeks of August 3 (27%), August 10 (25%) and August 17 (23%). That is exactly when back-to-school promotions peak. Retailers want a badge on everything a student might need, and the easiest way to get one is to raise the reference price, not lower the selling price. Office supplies and computers, the two categories most tied to back-to-school, were also the two with the highest fake-discount rates over the whole period (more on that in our category breakdown).

2. The big sale event was the most honest week

The report for the week of October 5 was taken on the Monday, the day before Prime Big Deal Days (October 6–7), when early event deals were already live. Fake-discount share fell to 8%, less than half the usual level. We also measured the event itself, once, while it was live: only 6% of the deals we checked were busted. Big events are when shoppers compare prices most carefully and when deal sites and reporters pay the most attention, so real price cuts are what sell. It is the quiet weeks in between, when nobody is checking, that carry the most padding. We wrote up the event in detail in Were Prime Big Deal Days 2026 deals real?

3. The rate snaps back after an event

After the event the share climbed again: 13% the following week, then 15%. It has not yet returned to the July–September level of around 20%, but the direction is clear. If you want the most honest pricing on a non-urgent purchase, the data says buy during a major event or check carefully before you buy in the weeks around it.

What “not really a deal” looks like

A busted deal usually looks completely normal. Here are real examples from the week of October 19, each shown with a discount badge on Amazon:

• A three-pack of Moleskine Cahier journals at $16.13. Its usual price over the previous 90 days was $14.41, so the “deal” was about 12% above normal.
• An ASUS NUC 14 Pro+ mini PC at $1,216.02, against a usual $1,098.83: about $117 more than it normally sells for.
• A Bosch SDS-Plus rotary hammer bit at $43.59, usually $39.04.

None of these is necessarily dishonest on paper. Each one may have some higher “was” price somewhere in its history or a list price set by the manufacturer. But if you bought any of them that week because of the badge, you paid more than you would have on an ordinary day. We go through more examples, and how to spot them yourself, in When the “deal” costs more than usual.

How these numbers are measured

Every week we take a fresh sample of advertised Amazon deals, about 1,500 to 1,750 of them, spread across 13 shopping categories so that no single category dominates. For each one we compare today’s price with the price that product has usually sold for over the previous 90 days, using our own price research. A deal counts as “not really a deal” (we call it busted) when today’s price is within 8% of that usual price or higher. In other words, the discount badge is there, but the price is not meaningfully lower than what the product normally costs.

That rule is deliberately generous to retailers. A product that is 9% below its usual price passes, even if its struck-through “was” price is wildly inflated. So the share you see is a floor, not a ceiling: the real number of misleading discounts is, if anything, higher. Each week’s report, its examples and a downloadable CSV are on the Fake Deal Index, and the full method is on our methodology page.

What to do with this as a shopper

• Ignore the percentage on the badge. It is measured against a price the seller chose. The only comparison that matters is today’s price against what the product usually costs.
• Be most careful in quiet weeks and during back-to-school. That is when the share of fake discounts was highest in our sample.
• Look at the category. Computers, office supplies and furniture carried the most padding; shoes and sporting goods the least.
• Check the price history before you pay. You can paste any Amazon link into worthitforyou and see where today’s price sits against a year of real prices. It is free, and no retailer pays for the result.

We publish the index every week. If you write about shopping or consumer prices, the numbers are free to quote with a link; see our press page.