How this works

We read the public catalogue of 424 shops every night, one request every 3.5 seconds, and record the price of every variant they list. That is 882,093 variants grouped into 3,400 products, across 18 countries and 10 currencies.

The last sweep finished 72 min ago. We have been recording since August 27, 2026 — 6 days of price history so far.

We do not scrape pages. Every shop here publishes its catalogue in a machine readable form of its own accord, and we read that.

A product is not a listing. The same board sold by six shops is one product with six offers, matched on the normalised title and verified on the URL — a path is written by each merchant independently, so when six of them write the same thing, that agreement is evidence rather than a tautology.

Every product then carries exactly one verdict, in this order:

• Real deal — cheapest of several independent shops, or a genuine discount no competitor beats. 653 today.
• Dropped today — the price fell between our last two readings of that shop. 550 today.
• New low — it came down to a price we had never measured, and is still there. 133 today.
• Seen lower — we watched it cheaper than it is now. 226.
• Best of N shops / Same price at N — several shops carry it. 1,222 and 884.
• Only listing — none of the 424 shops we read sells it anywhere else. 1,294, which is 38.1% of everything we track.

Only 2,106 products — 61.9% — are carried by two or more independent shops. Of those, 1,222 show a real price gap and 884 are the same price everywhere. We say so on both: an enforced price is information too.

• We never call a price the lowest we have seen until a full sweep has passed since we first read it. One reading proves nothing; two prove stability.
• We never convert currencies. Offers are compared inside one currency, never across. A converted comparison is a guess about a rate, not a measurement.
• We show the country a shop ships from. We do not know where it ships to, so we do not say.
• A crossed out price is only shown when a competitor genuinely charges it today. A discount a shop announces against itself is not a discount.
• Products we cannot compare are published anyway, and labelled as such. Hiding them would hide the one thing we are in a position to state.

Outgoing links pass through /go/ so we can count them, and some of them will carry an affiliate tag once a programme is signed. 304 of the products we compare have at least one merchant with a publicly announced programme.

No merchant pays for placement, and no verdict changes because of it. The order of a list is the order of the measurement.

The measurements are published under CC BY 4.0. Take them, cite the source. What is available and how to get it →