Prime Day, Measured
Prime Day came early this year, June 23 to 26, and Amazon's announcement promised millions of deals across more than 35 categories. Then, on July 14, the shelf data arrived. Omnia Retail, a Dutch pricing platform, tracked 1.19 million product prices across European retail through the event and published the tally. Of those prices, 74.6 percent did not move in any meaningful way. Another 11.6 percent went up. The share that received a genuine cut, more than 3 percent below the pre-event baseline, was 13.8 percent.
One price in seven. That is one of the most heavily marketed shopping events on the calendar, measured at the shelf instead of at the press release. I collect this kind of data for a living, so I want to walk through what else is in the dataset, because the scarcity of real discounts is not the interesting part. The interesting part is that none of these numbers exist unless somebody pointed an instrument at those shelves in early June and kept it there.
View data table
| Price movement | Share of prices |
|---|---|
| No meaningful change | 74.6% |
| Real cut (more than 3% below baseline) | 13.8% |
| Price increase | 11.6% |
Where the discounts actually went
Start with the 13.8 percent that were real. The median cut was 7.8 percent, the average 10.5, pulled up by roughly a quarter of discounters going deeper than 15 percent. The placement is the revealing bit. Products between 50 and 100 euros had the highest participation, at 15.6 percent. The deepest median cut, 8.7 percent, sat on products under 10 euros, where a dramatic percentage costs the least in absolute money. Above 1,000 euros, participation was the lowest of any bracket. Merchandising arithmetic, in other words: the drama goes where it costs least.
Then there is the part headline coverage skips. During a discount event, 11.6 percent of prices rose, and Omnia's series shows the hikers peaking at 11.1 percent above baseline right around Prime Day itself. Raising the reference price so the strike-through looks generous is the oldest move in retail promotion, and it is usually discussed as an anecdote about a single listing. Here it shows up in aggregate, as a measurable population of prices moving the wrong way at the exact moment attention peaks.
The competitive picture is my favorite detail. In the Netherlands, bol.com tracked Amazon's cuts closely on products they both sell, while MediaMarkt and Coolblue mostly held their prices flat and let the event pass. Three rational strategies for the same four days, and the divergence is only visible if you are watching all of them on the same clock.
Why none of this can be checked afterwards
A discount badge is a claim about the past: this used to cost more. Standing in front of the product page on June 27, you cannot audit that claim. The page shows today's price and whatever reference number the seller chose to print next to it. The only way to know whether 13.8 percent of a catalog was genuinely cut, rather than 80, is to have recorded the catalog before, during, and after the event, on a fixed schedule, from a consistent vantage point. Price history is not lying around waiting to be queried. It exists only where someone built the collection ahead of the question.
That is the actual product in my corner of the industry, and it took me years to learn to say it plainly. At Iceberg Data we run scheduled price collection across European retail catalogs, and the runs that earn their keep are the boring ones: the mid-June captures nobody was excited about at the time, which became the baselines every Prime Day claim gets divided by. Clients come in asking for today's prices. What they are actually buying is the denominator.
One more number makes the point about schedules. On June 30, the median price among products that had been genuinely discounted was still 8.3 percent below its pre-event baseline. Discounts decay slowly rather than snapping back. Sample only the event week and you overstate how concentrated the promotion was. Start sampling after it and you miss it entirely. Cadence decides what a price dataset is able to say.
For a consumer, there are tools that chart one product's history at a time, and they are enough to decide on a coffee machine. For anyone underwriting a retail position, the question is a portfolio one: how much promotional depth is really moving through European retail this summer, and whose margin absorbs it. This year's answer is 13.8 percent participation at 7.8 percent median depth. Whatever next summer's answer turns out to be, it is being collected right now, or it will not exist.