How often a “sale” price actually beats the typical price
The gap between advertised discounting and measured discounting, from our own observations.
Retailers advertise discounts constantly. We measure how many of them survive contact with the price a product actually sells for.
What we compare
For every offer we track, we hold two numbers:
- The claimed discount — the retailer's own sale price against their own reference price.
- The real discount — the same sale price against the median price we observed across every store carrying that product over the previous ninety days.
The first is a statement. The second is a measurement.
Current state of the dataset
| Measure | Current value |
|---|---|
| Products with both a claimed and a measured discount | 308 or more |
| Median overstatement where the claim exceeds reality | 12.0% |
| Median real discount among genuinely discounted products | 0.0% |
| Products carried by 3+ independent stores | 4 |
Why the gap exists
Reference pricing is unregulated in practice and easy to justify: a product genuinely did launch at a higher price, or is genuinely listed higher somewhere. Neither fact tells you whether today is a good day to buy. Only the distribution of prices actually charged does.
Caveats we will not paper over
This dataset covers independent specialist retailers, not marketplaces, so it is not a claim about retail in general. Products with short history are excluded from the real-discount calculation, because a median computed from two weeks of observations is not a baseline. And our coverage grows over time, so these figures will move — which is why they are queried live rather than frozen into the text.
How the real discount is computed
For each offer we take the effective price a buyer would pay — item price plus shipping where the retailer publishes it — and compare it against the median of every price we recorded for that same product across every store carrying it, over the preceding window.
Three rules keep that comparison honest.
Only matched variants count. Capacity, size, colour, condition, region and bundle contents must all agree before two listings feed the same median. A refurbished unit never contributes to a new unit's baseline, and a bundle never contributes to a bare item's.
Only fresh, in-stock, unflagged offers count. A listing outside our freshness window, out of stock, or carrying an unresolved price anomaly is excluded from the median rather than allowed to drag it.
Thin history is excluded rather than extrapolated. A product we have watched for a fortnight has no meaningful baseline, so it produces no real-discount figure at all. It shows its current cross-store spread instead — a measurement it does have.
What this does not measure
It does not measure whether a product is good, whether a retailer is trustworthy, or whether you should buy. It measures one narrow thing: how a price today compares to the prices that product has actually carried. That is deliberately less than most sites claim to tell you, and it is a claim we can support.
Related reading
Every figure in this article is read from UponSale’s live measurements as the page loads, not typed in by hand — so it cannot drift out of date. Read how we measure or where the data comes from.