How We Measure Prices and Stock

Some of the pages on https://www.vividrepairs.co.ukare built from our own measurements rather than from other people's reporting. This page explains where those numbers come from, what they cover, what they do not, and who checks them before they go out. If you want to quote one of them, this is the page that tells you whether you should.

1. What we sample, and how often

We track a catalogue of UK retail listings across PC components, laptops, peripherals and networking kit. Once a day, every active listing in that catalogue gets one reading: the advertised price, and whether it was in stock or out of stock at that moment. Each reading is stamped with the calendar day it belongs to, and one listing can contribute at most one reading per day.

The listings come from the retail sources we work with as an affiliate: Amazon UK, and the merchant feeds we receive through the AWIN and CJ networks. Every data story states which of those its own figures came from, with a count, rather than leaving you to assume.

2. What a "listing" is, and why it is not a product

We count listings, not products. The same graphics card sold under three different listings counts three times, because three separate prices moved. That is the honest unit for a pricing measurement and the wrong unit for a market-share one, so we never present these figures as market share.

Refurbished and renewed listings are excluded throughout. They price on a different curve and would drag every category median down without telling you anything about new stock.

3. Gaps, and why we would rather show you a shorter window

Our daily sampling stopped between 23 June and 7 August 2026, and those days are gone: no one can recover a price nobody recorded. Rather than paper over it, every story is built from the unbroken run of days ending at the most recent one, and states the two dates it actually covers. If a window looks shorter than you expected, that is why, and it widens by one day for every day the sampling runs.

For the same reason, we do not draw a line on a chart across a gap in sampling. A chart that connects June to August with a straight line is asserting six weeks of prices we never saw.

4. Minimum sample sizes

Every story enforces its own minimums and prints them on the page. Broadly: a listing has to have been read on several separate days before it counts at all, and a category has to carry at least thirty qualifying listings before it gets a row of its own. A category that does not clear the bar is left out rather than reported thinly, so an absent category means "not enough data", never "nothing happened".

Price movements are compared median to median rather than first reading to last, so a one-afternoon spike or a lightning deal caught on the final day cannot masquerade as a sustained change. Movements past a stated threshold are treated as listing errors until several readings on each side confirm them, and anything dropped that way is counted on the page.

5. Which listings we can name

Most of what we track is sold on Amazon, and our agreement with Amazon does not permit us to republish the historical price or stock record of an individual Amazon listing. Aggregate statistics across thousands of listings are a different thing, and those are what our data stories lead with.

Where a story names an individual listing next to its own price history, that listing comes from a retailer whose history we are free to publish. It is a smaller table than it could be, on purpose.

6. How the stories are written and checked

The figures are computed in code, directly from the readings, and are fixed before a word is written. The prose is then written around those figures, and an automated check reads the finished page and rejects it if it contains any number that is not in the underlying dataset. There is no path by which a figure on one of these pages can be an estimate, a rounding someone did in their head, or an invention.

Every story, including every refresh of one that has already published, is read and approved by a person before it goes live. Nothing on these pages publishes itself.

7. Corrections and the raw data

Each data story carries a download link to the complete dataset behind it as a CSV, including the method, the window, the sample size and the caveats. You are free to quote the figures with a link back.

If you think one of them is wrong, tell us at support@vividrepairs.co.uk and we will check it against the readings. If it is wrong we will correct the page and say what changed.

Related

• How we review, the research process behind a product review.
• How we score, what our ratings mean.