How we rate: methodology

We do not test units directly. We don’t own the robots we write about. RoboPool Clean Lab compares manufacturer specs and manuals, measurements published by named reviewers, owner reports and regulator recall records, and labels which is which. The word “lab” in our name means data analysis.

If we didn’t test it, why trust us? Because every number on the site says where it came from, and you can follow it back to the source. When the maker says one thing and a reviewer measured another, we show both.

Evidence labels

Every spec or performance number carries one label.

A maker figure repeated by a reviewer stays [C]. That includes a reviewer who lets a robot run for the time the manual states: the number is still the manual’s, not a timing. Reviews published by a business that sells the product or its rivals (for example, Poolbots) are retailer content, not independent testing. We label their figures [R], state the conflict, and never score or chart them as measurements.

Where our numbers come from

In order of weight:

1. Manufacturer manuals and spec PDFs (for specs only).
2. Regulator recall records: the US Consumer Product Safety Commission (CPSC) recall database, which we search directly, plus ACCC (productsafety.gov.au) and UK OPSS notices when we find one. We haven’t queried the EU Safety Gate portal yet.
3. Named reviewers who tested the robot themselves, cited by name and date.
4. Owner reviews, summarized in our words with the source and review count. This is not a systematic sample.

What we don’t use: expired-domain and parasite listicles, and brand blogs as proof that one product beats another.

How we record runtime

Runtime always states its mode. Eco, floor-only and “mode not stated” figures are flagged. When a maker’s own pages disagree, we show each figure and where it appears, and we don’t pick one silently. Example: Aiper’s own page for the Aiper Scuba S1 shows 150 minutes (Floor mode), 180 minutes (spec table) and 270 minutes (Enhanced Eco banner) [C].

Where no one has measured a model, the page says “no independent measurement found”. We never fill a gap with an estimate. When we calculate something ourselves, such as minutes per 1,000 sq ft, it is labeled “our calculation from [C] figures”. See the battery data page.

How we record prices

Every price shows the retailer and the date we checked it, read on the retailer’s own page. We refresh prices monthly and before Prime Day, Black Friday and the spring sales. A price older than 45 days is flagged on the page.

Pool-Fit Score in one minute

The Pool-Fit Score is a 0 to 100 fit check for one pool profile, not a quality grade. It uses six inputs: size headroom, wall and waterline coverage, runtime by mode against the pool, pool type and depth, filtration, and warranty. Missing data scores zero and reads “not published”. Full method.

Recalls

We search the CPSC recall database before each model page publishes and show the date of that search on the page. A recall is shown beside the score and never hidden. When we find none, we write “No recall found (searched {date})”, never “recall-free” or “safe”.

What we don’t claim

• No durability verdicts and no navigation verdicts from our own use, since we don’t use the robots.
• No star ratings and no scores out of 5.
• No “best overall” unless our published data ranks it, with the basis on the same screen.

Updates and corrections

We refresh money pages on a regular schedule, and every correction is logged on Corrections with the old figure, the new one and the source. To report an error, use Contact.

How we make money

As of launch, we have not joined any affiliate program. Commissions, if we ever earn them, never change our data or scores. See how we make money.

Keep reading: Pool-Fit Score · About · Corrections