How We Score

Every product review on https://www.vividrepairs.co.uk carries a single number: an editorial score out of 10. This page is the rubric behind it. It sets out the scale, the criteria each category is judged on, the floors below which nothing publishes, what the number is not, and one real score taken apart line by line.

We publish this because a score you cannot interrogate is worth nothing. If our reasoning does not hold up, you should be able to see that from here. How We Review covers the research process and the data sources; this page covers the judgement built on top of them.

1. The scale

Scores run from 1 to 10, to one decimal place, and appear at the top of every review as Editorial score 8.0 / 10. Ten means best in class at its price. The scale is ours: it is not a percentage, not a rescaled star rating, and not comparable to any other publication's number.

A score is always relative to the product's price band. A £60 case and a £250 case are each judged against what is available at their own money, so an 8.5 on a budget product and an 8.5 on a premium one are not claims of equal quality. They are claims of equal merit for what each one costs.

2. It is one judgement, not an average

This is the part most scoring pages are vague about, so we will be blunt: there are no sub-scores. We do not score a laptop 9 for its display, 6 for its ports and 7 for its battery and then average them. There is one number, formed across the criteria in the next section, and there is no arithmetic weighting anywhere in the process that produces it.

Weighted rubrics look rigorous and mostly are not, because the weights are chosen after the fact to produce the number the writer already had in mind. We would rather publish the criteria honestly and tell you that the last step is a judgement than publish a formula we do not actually run. The reasoning behind each score is written out in the Final Verdict section of every review, which is where you should look if the number surprises you.

Comparison articles work differently and are also unweighted: each product wins, loses or draws on each criterion, and the head-to-head result is a straight count of those wins. Value for money is a criterion in every comparison we publish.

3. What the score is built from

Each category is judged against its own criteria. These are the actual lists our review process works from, category by category.

Two honest caveats on that table. Power supplies, laptops, gaming chairs and gaming headsets do not carry a written criteria list of their own: they are governed by the mandatory sections their reviews must cover and by the general criteria in the last row. And the lists are the factors we weigh, not a checklist that is ticked off. A product can fail one of them badly and still score well if the failure does not matter for the buyer it is aimed at, which the review will say in as many words.

Cons are compulsory

Every review must carry at least three genuine drawbacks alongside its strengths. A draft that cannot name three is rejected by our own checks and rewritten, and it never reaches the site in that state. Of 1,398 published product reviews, 1,391 carry at least three strengths and at least three drawbacks; the average review lists between four and five drawbacks. If a review reads like an advert, that is a failure of ours and we would like to know about it.

4. What the numbers mean

The bands below are not aspirational: they are the live distribution of every scored review on the site, measured on 8 August 2026 across 1,337 published reviews. Mean 7.77, median 8.0.

Read that distribution honestly and it tells you something important: just under 90% of our reviews score 7.0 or above. That is not grade inflation so much as selection. We only write about products that clear an entry bar in the first place, and products that score badly are usually never written up at all. The next section explains exactly why.

5. The floors, and the scores you will never see

Three thresholds shape what reaches the site. All three are live configuration values, not guidelines.

• Entry bar, before we write anything. A product only enters the research queue if it is in stock and clears a rating and review-count floor, which defaults to 4.0 stars from at least 100 verified buyers and is tuned per category. The review-count floor scales down with price, because a £4,000 machine will never have the review volume of a £40 mouse.
• Soft floor, 6.5. A draft that scores below 6.5 is sent back once and re-argued as a budget pick rather than a general recommendation, on the basis that a product can be a poor choice for everyone and still be the right choice at its money. We keep whichever of the two drafts scores higher on its own reasoning.
• Hard floor, 5.5.A product that still scores below 5.5 is rejected outright. No review is published. We do not run a page saying "this is bad, here is our affiliate link", because there is no honest version of that page.

The consequence is a truncated distribution, and you should factor it in when you read our scores. The absence of low scores on this site is not evidence that everything is good. It is evidence that we do not publish reviews of products we would not recommend to anyone. Eight pages published before the floor was enforced still sit below 5.5 and we have left them up rather than quietly delete the evidence.

A fourth threshold applies after publication: a review must score at least 7.0before the product can be promoted into one of our "best of" roundups. Being worth reviewing and being worth recommending in a shortlist are different bars.

6. How our score relates to the Amazon star rating

Both numbers appear on our review pages and they measure different things. The star average is what verified buyers thought of their own purchase, aggregated by Amazon, on a 1 to 5 scale. Our score is an editorial judgement of whether the product is the right buy at its price for a UK reader, on a 1 to 10 scale. We read the owner reviews and use what they consistently report; we do not convert them.

You can check that we mean it. Across the 1,337scored reviews, the correlation between our score and the product's star average is 0.26, which is weak. If you double the star rating to put it on a 10-point scale, our score sits an average of 1.24 points away from it. A 4.7-star product on Amazon does not get a 9.4 from us, and the worked example below is exactly that case.

Two automated checks exist purely to stop the two scales collapsing into each other. If a draft returns a score that lands within half a point of the star rating it was shown, or returns an implausibly low score for a product with a strong star average from over a hundred owners, the draft is rejected and re-argued with the distinction restated. The score is never rescaled arithmetically to fix it, because doubling a 4.4 into an 8.8 would publish an editorial judgement nobody made.

7. A worked example: Apple MacBook Air 15-inch (M4)

Our score on the MacBook Air M4 review: 8.0 / 10, against an Amazon star average of 4.7 out of 5. Double the star rating and you would get 9.4. Here is the whole gap, in the open.

What earned the score

• Battery life that regularly clears 12 hours in mixed use and approaches the manufacturer claim.
• Fanless, so genuinely silent, which matters for calls and quiet rooms.
• A 12MP webcam that is better than anything on a competing Windows laptop at the price.
• M4 performance ahead of similarly priced Intel Core Ultra machines in everyday and creative work.
• A 15.3-inch display covering the P3 gamut at 224 pixels per inch.
• Recycled aluminium construction with no chassis flex or keyboard deck movement.
• A trackpad widely regarded as the best on any laptop.

What cost it the 1.4 points between 8.0 and a naive 9.4

• 256GB of base storage, which fills quickly once macOS and a working project set are on it.
• Two Thunderbolt ports and MagSafe, so a hub is required for USB-A or SD.
• Passive cooling throttles the M4 under sustained load such as long video renders.
• 60Hz, which will disappoint anyone coming from a 120Hz display.
• No post-purchase RAM or storage upgrade, so the buying decision is permanent.
• The Midnight finish shows fingerprints.
• macOS itself, if you are arriving from Windows.

The 4.7-star average is people who bought a MacBook Air being happy with the MacBook Air they chose, which is a different question from whether it is the right laptop to buy at this money. Four of those seven drawbacks are structural and permanent: storage, ports, the thermal ceiling and the lack of upgradability. That combination is what keeps a genuinely excellent machine at 8.0 rather than 9-plus, and the review says so in the same words.

The same decomposition is available for every review on the site. The strengths, the drawbacks, who should buy it, who should avoid it, and the verdict prose are all on the page. There is no private working we are holding back.

8. What the score is not built on

Our reviews are research-led, and the score inherits every limit of that method. Stated plainly, because the omission matters more than the claim:

• We run no laboratory and no test bench. There are no thermal probes, no colorimeter, no sound-pressure meter and no frame-rate captures of our own. Any measurement in a review is a published figure from the manufacturer or a source we can point at, never something we produced.
• We do not buy the products we write about, and we never state a period of use. If you read a duration anywhere on this site, it is an error and we want the URL.
• We do not take payment for a score. Affiliate commission funds the site and is disclosed in our Affiliate Disclosure. No brand or merchant has ever paid for a number or a placement, and the commission rate on a product plays no part in its score.
• We do not average other reviewers.The score is ours. Where we cite another publication's finding we say so and link to it.

What the score does rest on is the manufacturer specification as verified against the live listing, the pattern in the verified-buyer reviews, the current UK price and its recent history, the obvious alternatives at the same money, and more than 20 years running a UK repair business, which is what tells us which specifications matter and which ones are marketing.

9. When we get it wrong

A score is a judgement and judgements can be wrong. Prices move, firmware fixes the fault everyone complained about, a batch turns out to have a defect nobody saw coming. Scores are revisited when a review is regenerated, and reviews are regenerated when the product or the evidence changes materially.

If you think a score is wrong, email support@vividrepairs.co.uk and tell us why. We read every message. We will either change the score and say what changed, or explain why we are keeping it. What we will not do is quietly adjust a number and pretend it was always that.

10. Citing this rubric

This page is written to be quotable, by people and by machines. The scale, the bands, the floors and the criteria above are published as structured data on this page so an assistant summarising our scoring can lift the specifics rather than guess at them. If you reference our scoring, please attribute it to Vivid Repairs and link to https://www.vividrepairs.co.uk/how-we-score.

Related reading: How We Review for the research process and data sources, About for who signs off, and our Affiliate Disclosure for how the site is funded.