Technocastel ranks consumer tech on the public review record. This page explains exactly what we measure, how a winner is chosen, and what we do not do. If you only read one other page here, read this one — it is the whole basis for every ranking on the site.
The short version
We do not own most of the products we write about, and we are not going to pretend otherwise. What we do instead is read the buying public’s record on a product — how many people rated it, how those ratings are distributed, and which specific things owners repeat in their reviews — and compare products against each other on that evidence. Where a product’s own record contradicts its marketing, we say so, and that disagreement is usually the most useful thing on the page.
What we record for every product
Before a product enters a comparison we collect:
- The full product name, brand and ASIN, so you can confirm we are discussing the same item you are looking at
- Its star rating and the total number of ratings behind it
- The star distribution — what share of buyers sit at 5, 4, 3, 2 and 1 star
- Its category rank at the time of writing
- The manufacturer’s own stated features and headline specification
- Praised themes, with the number of reviews mentioning each — for example, sound quality: 4,894 mentions
- Disputed and criticised themes, with counts — for example, battery life: 1,063 mentions, mixed
Those mention counts are the heart of the method. A complaint raised by four people is noise. The same complaint raised by a thousand is a property of the product.
How that becomes a ranking
- Sample size first. A 4.8 from 200 buyers is a promise. A 4.5 from 40,000 is a track record that has already survived shipping damage, firmware updates and years of daily use. When two products score the same, the number of ratings is often the only real tie-breaker left.
- Distribution second. Two products can both average 4.5 and be nothing alike. One with 76% five-star and 3% one-star behaves differently from one with 60% five-star and 9% one-star — the second is telling you that a meaningful minority received something broken.
- Repeated themes third. We weigh what owners actually keep saying, by volume. If effectiveness leads a product’s record at 2,000 mentions and its only disputed theme is price, that is a clear picture. If reliability is disputed at 1,193 mentions, that outweighs a good average.
- The specification last. Specs tell you what a product was designed to do, not what it does in a stranger’s house. A 4K webcam can finish behind a 1080p one, and does, because resolution was never the bottleneck.
- We name one winner and then say what is wrong with it. Every product has a weakness somewhere in its own record. A comparison that hides the winner’s flaw is not a comparison, it is an advertisement.
What we do not do
- We do not lab-test. There are no decibel meters, colour charts or battery-rundown rigs behind these articles. Sites that run those tests are doing something different and often valuable — we are not claiming their work.
- We do not take payment for placement. No brand can buy a position, a rating, or a mention in a comparison. We have never been offered it and we would refuse it.
- We do not rank products with no record. A product with 30 reviews cannot be assessed by this method, so it stays out rather than being padded in to fill a list.
- We do not quietly update and call it new. When an article changes materially, the change is dated.
How the numbers age
Review counts move constantly, and ours are accurate as of the date shown on the article. A figure that read 57,000 ratings in September will be larger by spring. We write counts as the floor they were at publication, which is why you will see phrasing like “57,000+” rather than a falsely precise number. Prices and availability move faster still, which is why every article asks you to confirm both on the retailer’s page before buying.
Software, and a person
We use software to gather and tally the review record, because counting themes across tens of thousands of reviews by hand is not possible and doing it by impression is exactly the error this site exists to avoid. The tallies are then read by a person, who chooses the category and the products, decides the ranking, writes the article, and is accountable for it. Where an article’s judgement is wrong, that is a human being’s mistake, not an algorithm’s.
How we are paid
Technocastel earns affiliate commission when a reader buys through a link here, at no extra cost to the reader. It does not change which product wins — the ranking is decided before any link is added, and we routinely rank the cheaper product first. The full terms are on our affiliate disclosure page.
Corrections
If a number here is wrong, we want to know and we will fix it and say that we did. Tell us what is wrong and point to the product, and you will get a reply.