Editorial method

How we review

How we search the web, filter useful evidence, choose one decision-helping piece, and turn the result into a score.

Last updated September 29, 2026

What this is

99deals Reviews searches the public record on a product before you have to. We check the brand's claims and manuals, Amazon and other retailers, YouTube, public Instagram posts, Reddit and specialist forums, and relevant editorial or regulatory coverage. Then we filter what we found down to the evidence that can change a buying decision.

Every sentence on a review page that carries a number or a judgment about performance names its source in the same sentence. If we could not attribute a claim, it does not appear on the page.

Where the evidence comes from

Each review lists its research footprint: the volume of buyer ratings and reviews represented by the sources we checked, plus the videos, public social posts, listings, and forum threads that qualified. A represented count does not mean we read every review in that count. When a marketplace blocks a useful check, we record the limitation instead of filling the gap with an estimate.

Sources are named on the page, not linked. Two reasons: a link out to a store is a commercial act rather than a citation, and the pages we cite change. The full record with URLs lives in the working file behind each review.

When a product is too new or too obscure to have its own record, we say what we found about the class of product and the same item sold under other names, we label the review as class scope, and we cap the score it can earn. A product never inherits reviews it does not have.

Google helps us discover pages, but a search result is never evidence. Searches that produce no useful result are recorded too. That lets the review show that a channel was checked without pretending it added something to the decision.

How we choose the featured piece

Each review includes one piece that is most useful for deciding. It may be a YouTube review, a summarized public Instagram finding, or a brand demonstration that clearly shows use, setup, or specifications. Instagram findings are summarized without copying the post or linking out.

Selection factorWeight
Usefulness for the buying decision35%
Exact product match and demonstration25%
Evidence that can be corroborated20%
Clarity10%
Recency5%
Reach or engagement5%

Views are context, not a threshold. A new video can be the better choice if it provides better evidence. Any material relationship tied to a source is recorded during research and limits how much weight that piece can carry. No selected piece supports a performance claim by itself.

How the score works

Five criteria, fixed weights, 0 to 10 each. The total is the weighted average, and each criterion carries one line saying what the evidence was.

A brand assessment is different from a product score. It can show an equal-weight average of the reviewed products plus a separate manual brand-context judgment, but only after the site owner explicitly approves it. When that happens, the brand page shows the calculation and reason. The adjustment never enters a product score, criterion or verdict.

Criterion Weight What it measures
Solves the problem 25% Whether buyers report it doing the job it is sold to do, and how often it falls short.
Quality and durability 25% What owners say breaks, wears out or holds up, and how long after purchase they say it.
Fit and ease of use 20% Sizing accuracy, learning curve, and the friction between opening the box and using it.
Total value 20% What it costs against what comparable products cost and deliver.
Care and design 10% Cleaning, storage, materials and the design choices that show up in daily use.

What each verdict means

Verdict Score What it means
Worth trying 7.0 – 10 Enough evidence that it does the job, with a stated condition and what to check when it arrives.
Consider it 5.0 – 6.9 It may fit a specific buyer. We state the condition and the tradeoff to weigh.
Compare first Below 5.0 Another option may fit the job better. We show what deserves comparison, and you decide.

A safety signal or a regulatory problem overrides the arithmetic. Where a gate and the score disagree, the more cautious of the two guides the page. A safety or regulatory flag is not published as a buying invitation; it is blocked for review instead.

These are orientations, not orders. We do not make the choice for you. We show who a product may fit, what to weigh, and what deserves comparison next.

Our score vs. customer ratings

These are two different things and we never merge them. Our score is editorial, on a scale of 10, and one desk produced it. A customer rating is on a scale of 5, belongs to whoever collected it, and always appears next to the name of the site it came from and the date we read it.

We do not publish a combined or overall star average, and we do not mark one up for search engines. Averaging a rating we did not collect would be claiming a sample we do not have.

How we make money

99deals Reviews is published by 99deals.shop, an online store. That relationship is stated in the header of every page on this site, and it shapes how links work here:

  • If we sell the product, every purchase button goes directly to its page in our own store.
  • If we do not sell it, every purchase button goes directly to one verified retailer product page. That destination may earn us a commission, disclosed in the purchase section.
  • Products we mention only for comparison carry no link at all.

A purchase destination may appear near the verdict, in the decision summary, and at the end of the review. Those buttons repeat the same verified URL so a shopper never has to pass through an intermediate section before reaching the product.

Some products we review are made by companies that are clients of our parent business. When that is the case, the review says so on the page. The commercial relationship does not reach the scorecard: criteria are filled in from the evidence before any link is placed.

See commercial relationships for the full statement.

Getting it wrong

Prices move, listings change, and manufacturers reformulate products without renaming them. Every review carries the date its research was checked and the date it is next due for a refresh. When something we published turns out to be wrong, we change the page and note what changed rather than quietly editing it.

If you find an error, tell us.