Robot vacuums
Hard floors · Carpet and rugs · Pet hair · Mopping · Navigation and obstacles · Dock and upkeep · App and software · Noise · Reliability and support · Value
How we compare
We don't score robots.
We report what reviewers found.
We haven't tested these robots ourselves, so we don't give them a score. We gather independent reviews, report what they found area by area, and are open about which robots the evidence favours.
What counts
A qualifying review.
A review counts only if it is independent (not the maker, a retailer, or a sponsored or affiliate-led piece), hands-on (someone actually used the robot, rather than restating the spec sheet) and finished (not a first look or a test still in progress). A disclosed free review sample is fine.
Reviews that miss one of these tests can still add detail, and we say so beside them, but they never decide a verdict, a badge or a pick. A robot with fewer than two qualifying reviews gets a basic page marked “Limited evidence”.
What reviewers found
A verdict for each area, in words.
Every full profile shows how reviewers judged each area that matters for its category. The verdicts are words, never numbers, stars or averages, and each one names the reviewers behind it with a link to their review.
- 01
Reviewers rate it highly
At least two qualifying reviews covered this area and were positive.
- 02
Mixed results
Reviewers disagreed, or found it only adequate. We explain the conditions behind the difference, such as thick carpet or a sloping lawn.
- 03
Reviewers found problems
At least two qualifying reviews covered this area and found it weak.
- 04
Not enough evidence
Fewer than two qualifying reviews covered it. If one did, we say what it found and name it.
The areas we report on
What we look for, by category.
Robot lawn mowers
Cut quality and edges · Setup and boundaries · Navigation · Slopes and rough ground · Obstacles and safety · Weather · App and software · Upkeep · Reliability and support · Value
Robotic pool cleaners
Floor, wall and waterline · Debris · Coverage and navigation · Battery and charging · Filter and retrieval · Pool types it suits · Reliability and support · Value
Window-cleaning robots
Cleaning result · Safety · Surfaces it suits · Ease of use · Noise · Value
Companion robots and robot dogs
What it really does · Staying engaging · Who it suits · App, subscription and running costs · Privacy · Build quality and battery life · Support and longevity · Value
Humanoid robots
Who can buy it · What it can do today · Safety · Software and platform · Support and running costs
Areas that don't apply to a robot, such as mopping on a vacuum without a mop, are left out.
Badges and picks
Which robots the evidence favours.
Badges. A full profile shows either “Highly rated by reviewers” or “Independently reviewed”; anything less is “Limited evidence”.
Our picks. Buying guides name picks such as “Our pick for pet hair”. A pick must have a full profile, the strongest verdicts in the areas that matter for that need, and a confirmed price where you're shopping. Each comes with a “Why we picked it” line naming the reviewers. Guides rank robots by the same rules, never by a score. A pick is our preference based on reviewers' work, not a test of our own.
Other scores and ratings. A profile can show a reviewer's own score, one reviewer at a time and named, such as “TechRadar: 4.5/5”. We never average or combine them. Customer star ratings stay separate and say which shop they come from and how many ratings there are.
Buying links
Useful routes to the exact product.
Links should point to the correct regional variant with clear price, availability, warranty and returns information.
- Exact model and variant
- Correct regional store
- Current price when available
- Paid links clearly labelled
Comparison questions
Clear answers.
When does a robot get a full profile and a badge?
When at least two qualifying reviews exist: independent, hands-on and finished. Robots below that bar still get a page with their specs, features, prices and who they suit, marked “Limited evidence”; the badge appears once a second qualifying review exists. Where a reviewer tested a sibling model, we say which one, and only use what the models share.
What does “Highly rated by reviewers” mean?
It needs a full profile (at least two qualifying reviews). Reviewers must have covered at least three of the main areas for its category, such as cleaning, cutting or safety (or all of them, where a category has fewer than three). More than half of the areas they covered must be rated highly, and none can have problems. An area with not enough evidence doesn't count against a robot. Every other robot with a full profile is “Independently reviewed”.
Do you test robots yourselves?
Not yet. Everything on a profile comes from independent reviews, the maker's specifications and owners' reports, and we name where each finding comes from. When we test robots ourselves, we'll say so on the page.
Why don't you show stars in search results?
Because the research we gather isn't a set of reviews submitted to us, and we don't score robots ourselves. We don't add review-summary star markup to any page.
Can a strong product still be a poor choice?
Yes. A capable robot can still be poor value, unavailable in your country or unsuitable for your floors, lawn, pool or windows.
Why can verdicts and picks change?
New reviews, firmware updates, prices and availability all change. Verdicts are revised when a new qualifying review arrives, and picks follow the verdicts and local prices.