Owner-review methodology

Public review statistics, paraphrased patterns, source links.

Manufacturer marketing tells you what a tap should be. Owner reviews tell you what owners actually report after eighteen months in a kitchen. We summarise the publicly visible aggregate statistics and link back to the source so you can read the full reviews yourself.

We do not republish review text. We do not run automated scraping. We do not claim our aggregation is the source platform's data. What you see here is our own statistical summary of public information, with every figure linked to the underlying source.

What we do

Collect only publicly visible aggregate statistics: total review counts, average ratings, star distributions.
Paraphrase recurring themes (top pros and cons) in our own words. Never republish review text verbatim.
Link to the source platform on every aggregation so you can read the underlying reviews yourself.
Refresh manually on a rolling six-week cycle.
Disclose the last refresh date on every model page so you know how current the data is.

What we do not do

Reproduce individual review text verbatim, in bulk or otherwise. Copyright sits with the original reviewer.
Run automated scraping bots against review platforms or violate any platform's terms of service.
Use review platform data to claim our aggregation IS that platform's verified data. Our summary is our own analysis OF public data.
Filter or weight in a way that favours commercial partners. Methodology is identical for every model.
Publish without source attribution. Every aggregation links to the platform of origin.

Where we look and what we record

Trustpilot

What we record: Aggregate brand rating, total review count, 5-star and 1-2-star distribution. We do not reproduce individual review text.

Best signal for brand-level service quality and after-sales handling. Linked source so you can read full reviews yourself.

Source: https://uk.trustpilot.com

Amazon UK

What we record: Product-level average rating, verified-purchase review count, star distribution.

Strong for product-level patterns. We filter to verified-purchase counts only where the platform exposes them.

Source: https://www.amazon.co.uk

Currys

What we record: Aggregate Reevoo rating + total review count per product.

UK-specific buyer signal. Smaller sample size but high-trust verification.

Source: https://www.currys.co.uk

John Lewis

What we record: Aggregate star rating + review count per product.

Strong UK premium-buyer signal. Useful for warranty-experience patterns.

Source: https://www.johnlewis.com

Manufacturer site

What we record: Public review counts and ratings where displayed.

Lowest verifiability tier. Included only for completeness and always cross-referenced against third-party sources.

Quality filters

A raw review count is meaningless without filtering. Here is what we apply before we publish an aggregate stat.

Only verified-purchase reviews where the platform exposes verification status.
Aggregate from at least 50 reviews per model before we display a confidence-weighted rating.
Themes are surfaced only if they appear in at least eight independent reviews (paraphrased in our own words).
Anything below the threshold is flagged as anecdotal or hidden until the sample size grows.
Reviews older than 24 months are de-weighted because the product version may have changed.

Individual reviews are copyrighted by the reviewer who wrote them. We do not reproduce review text in bulk. The aggregate statistics we publish (counts, averages, distributions) are factual data. The thematic summaries we publish are our own paraphrases in our own words. Where a single owner quote is materially illustrative we use it sparingly, under fifteen words, in quotation marks and with attribution to the source platform. If you are a review platform or a reviewer and you believe a specific aggregation is inaccurate or infringing, contact the editorial team at info at nextdayonline.nl and we will respond within five working days.

Coverage status

Our review aggregation database is being built. We target a 50-review minimum per model before we display a confidence-weighted score on the model page. Models currently below that threshold show the source links so you can browse the reviews directly while our coverage grows.