What costs less today than it did years ago
1,592 products we track are cheaper today, in plain dollars, than a dated capture of the shop's own page shows them costing years ago. The median comparison reaches back 3.6 years, and the median product costs 23% less than its own archived capture.
The comparison, and the link that proves it
Each row below is a product whose price today is lower than a dated snapshot of the same shop's own page. The last column links to that snapshot on the Internet Archive. Nothing on this page asks to be believed: open the capture and read the price yourself.
The live list, rebuilt twice a day, is at cheaper than before.
The two guards that decide what qualifies
An old price that is barely higher than today's proves nothing — catalogues round, fold shipping in, and drift. Two thresholds keep this page from measuring noise:
- The old price must exceed today's by at least 10%. Under that, the difference is indistinguishable from the ordinary churn of a catalogue, and calling it a four-year-old saving would be dressing up a rounding error.
- The capture must be at least two years old. Below two years the sentence collapses: “cheaper than last year” is a gap our own twice-daily record will soon measure by itself, and the Archive would add nothing that we could not see ourselves. Two years of inflation, by contrast, are real, and “cheaper than 2024” means something.
We take the oldest capture that clears the bar, not the most expensive one. What carries the claim is reach, not spread — a $900 fall measured against last spring is a weaker statement than a $200 fall measured against 2019.
What an archived price is allowed to do
This is the rule that governs every use of the Archive on this site, and it runs in one direction only: an archived price can take a claim away, or date one that holds. It can never grant one.
Our own readings begin on 27 August 2026. When we say a price is the lowest we have ever measured, that is a statement about our record and nothing more. If a dated capture shows the same page cheaper in 2020, the claim is simply false, and we drop it — the fact that it was true in our log does not save it. That is the entire reason we went and read twelve years of captures: not to find things to boast about, but to find out where we would otherwise have been wrong.
What an archived price never does is create a verdict. A capture cannot make something a deal, because we did not read it, we cannot re-read it, and a page photographed once in 2021 tells us nothing about whether it could be bought that day. A price captured on an out-of-stock page is never treated as a low at all: the page was showing a number nobody could pay.
The spread
How far apart the two prices sit, across all 1,592 products, in each shop's own currency — which means these figures cannot be added together, only read as a shape.
| Cheaper by | Gap in the shop’s own currency | |
|---|---|---|
| Smallest that qualifies | 9% | 1 |
| Lower quartile | 15% | 15 |
| Median | 23% | 39 |
| Upper quartile | 35% | 151 |
| Largest | 90% | 2290 |
The percentage is the column to read. The amounts beside it are in eleven different currencies and cannot be added, averaged or compared against each other — a median of “39” names no sum of money. The percentage survives the mixture intact, which is why it is the figure we lead with.
Read against what prices actually do over time, this page is the minority case: most pages never move, and of those that do, more rise than fall. These are the ones that went the other way, and every one of them comes with its receipt.
Questions this raises
How do you know what something cost years ago?
From a dated capture of the shop's own product page held by the Internet Archive. Every row links to the capture it came from, so the claim can be checked without trusting us. We hold enough of these to document 1,592 products today.
Why must the old price be at least 10% higher?
Below that, the difference is indistinguishable from the ordinary churn of a catalogue — rounding, shipping folded in, a supplier change. Calling that a four-year-old saving would be dressing up a rounding error.
Why two years, and not one?
Under two years the sentence collapses. “Cheaper than last year” is a gap our own twice-daily record will soon measure by itself, and the archive would be adding nothing we could not see for ourselves. Two years of inflation are real, and “cheaper than 2024” means something.
Can an archived price make something a deal?
Never. An archived price can take a claim away, or date one that holds, and nothing else. We did not read it, we cannot re-read it, and a page photographed once tells us nothing about whether it could be bought that day. A price captured while the page said out of stock is not treated as a low at all.
Why go to the trouble of reading twelve years of captures?
Not to find things to boast about — to find out where we would otherwise have been wrong. Our own readings begin on 27 August 2026, so “the lowest we have ever seen” is a statement about a very short record. Where a dated capture shows the same page cheaper in 2020, we drop the claim.