We run fazy.com, a deals site that records what products actually sold for. Last week we asked our own data a simple question: when a retailer shows a "was" price, was that ever the price?
For 83% of products, the answer was no. Not once in our tracking history did the product sell within 2% of its "was" price.
How we measured: every product in our catalog showing a reference price at least 5% above its current price, price read within the previous three days, tracked for at least 30 days with 4+ observations. Average watch time: eight weeks, 13 observations each.
By category, the worst offenders: wellness 97%, cosmetics 96%, home 91%, footwear 91%, electronics 88%. Apparel was the most honest at 68%.
43% of products showed the same price at every single check. A crossed-out price that never moves is not a discount. It is the presentation.
Full methodology: https://fazy.com/blog/was-prices-we-never-saw
Why I am posting this here: the dataset is queryable. We expose a free, no-key deals API and MCP server with price history: https://fazy.com/developers and https://fazy.com/mcp. Any agent can ask "what did this actually cost over the last 90 days" instead of trusting the tag.
I am the founder of Fazy, so read the numbers skeptically and check the methodology. That is the point of publishing it.
Top comments (0)