In a shop, the price on the shelf is the same for everyone standing in front of it. Online, it might not be. ShelfPrice measures what people are actually shown — and publishes only the differences that survive its own exclusions.
Free. No account. Open source. It never records the pages you visit.
Personalised pricing is individual by design. From where you are standing there is nothing to compare against, so it is invisible — which is exactly why it works.
A scraper has one address and one profile, so it collects its own price over and over. You cannot detect a difference between people by watching from a single vantage point.
Enough independent observers on the same item, and the distribution becomes visible. That is the only way to see it, and it is the reason a competitor cannot copy this with a crawler.
Currency. Shipping. Tax. Region. Stock. A live A/B test. Any of these produces a difference that has nothing to do with who you are — and a chart that ignores them looks dramatic and falls apart the first time someone checks it.
So the only claim ShelfPrice will ever publish has this shape:
If we cannot say it in that shape, it is not a finding and we do not publish it. The method and its thresholds are written down and versioned before the data is collected, and they are never changed after seeing a result.
Observations from real shoppers, used to decide where to look. They rank which retailers and items are worth measuring properly.
They never become a published number, and they are never shown to you as a verdict. That also means a faked report can waste our time but cannot enter the record.
A controlled measurement we run ourselves: the same item, the same moment, across several profiles and vantage points, with the boring explanations pinned and excluded.
Every published figure comes from here, and names the method version it was measured under.
So the guarantees are properties of the code, not promises in a policy. You can read all of it.
| What | How it is guaranteed |
|---|---|
| No account, ever | There is no login and no user record. Nothing to breach, nothing to hand over. |
| Your browsing is not recorded | The extension stores a product identifier. Any observation containing a web address, a search term, a location or a cookie value is refused, not cleaned up. |
| It does not run in the background | There is no always-on page script. It reads a page only while you have the popup open on it. |
| Observations cannot be chained into a history | The random identifier rotates every day, so even we cannot join yesterday to today. |
| Your location stays coarse | Postcodes are cut to three characters. No precise location, no IP address — the server never logs one. |
| You can see and delete everything | Observations sit on your device until you export or delete them. |
Where a shopping tool usually makes money is a commission on the "buy it cheaper here" button. That quietly decides which shop you are sent to, and it has ended other tools' reputations.
ShelfPrice takes no commission and never will. There is no referral code in the code and no place to add one. Donations and at-cost merch, later. Researchers are never charged.
The extension works and shows you your own price broken down. The comparison across people needs enough observers on the same item before it can say anything, so it stays quiet until it can.
No findings have been published yet. When they are, they will carry their method version and the exclusions that were applied.
One permission, once, when you install it. After that it never asks again — and you can switch any individual site off from the popup without another prompt.
Chrome Web Store listing to follow. Until then it loads unpacked from the source.