How we pick
We have not tested any of these machines. Here is what we do instead, stated in enough detail that you can check it.
We do not test robot vacuums. We have never owned one of the machines on this site. What we do is compile published specifications from manufacturer pages and listings, record the source URL for every individual figure, mark the gaps as gaps, and rank on criteria stated before any product is named.
What we do not do, first
It matters that this comes first rather than in a footnote. We have no test lab, no purchased units, no floor rig and no engineer. Anyone telling you they lab-tested twenty robot vacuums this year either has a lab or is not telling you the truth, and we have neither the lab nor the appetite for the second option.
So you will never read the words in our testing, we found, we ran it over a 40mm threshold, or in real-world use anywhere on this site. Those phrases describe contact with a machine, and we have not had any. Where genuine hands-on evidence exists — Vacuum Wars runs real threshold tests, for instance — we cite it by name and link to it, as someone else's work rather than absorbing it as ours.
What we do instead, in steps you could repeat
- Pull the published specification from the manufacturer's own page where one is reachable: suction in Pa, threshold climb height in mm, mop lift height in mm, brush construction, navigation system, stated runtime, dustbin capacity, body height, dock behavior.
- Fall back to the manufacturer's own retail listing where the product page is unreadable or silent, and label it as a listing rather than a spec page. Manufacturer pages outrank listings; listings outrank review sites; we do not use review sites for numbers at all.
- Record the source URL for every individual figure, not one blanket citation per product. Every number on this site is a link.
- Record the retrieval date. Ours is 2026-09-01 for this build. Specifications change and a number without a date is a number you cannot audit.
- Normalize units so cross-brand comparison is valid, and show both mm and inches on physical measurements because manufacturers split between them.
- Where a manufacturer does not publish a figure, record it as not published. We never substitute a competitor's number, a review site's estimate or a guess. The blanks in our tables are findings, and several of them are the most interesting thing on the page.
- State the condition next to the number where the maker states one. Dreame's 6cm climb is 4.2cm for a single step; Roborock's 4cm is a double-layer threshold; ECOVACS' 20mm is with the mop attached. A headline figure without its condition is close to useless.
How picks are ranked
Every roundup states its criteria before it names a product, in a How we picked block on the page itself. The criteria differ by page because the problems differ — a pet-hair roundup ranks on brush design, a threshold page ranks on published climb height. What does not differ is that the criteria are written down first and the products are fitted to them, rather than the other way round.
General principles that apply everywhere:
- A published figure beats an unpublished one. A maker that states its climb height ranks above one that markets chassis-lift hardware and states nothing. That is a judgment about transparency and we apply it consistently.
- Conditions count against headline numbers. A conditional 60mm and an unconditional 40mm are not ranked as 60 versus 40.
- Fit for the stated problem beats overall quality. The best machine on this site is not the top pick on every page, and should not be.
- Price is not a criterion we can rank on, because we render live prices only and never a remembered one. Where a pick is chosen partly for costing less, we say so in words.
What disqualifies a product
- It is not currently sold new in the US market.
- The manufacturer publishes nothing verifiable about it — a product we can say nothing sourced about is a product we cannot rank.
- It is a renewed, refurbished or bundled listing rather than the product itself.
- It is sold under a brand with no traceable manufacturer or support presence.
Why every roundup names a pick to skip
A list where everything is recommended is a list that has not made a judgment. Every roundup on this site names at least one product to skip and says why, including products we recommend elsewhere for a different problem. The skip entry is often the most useful paragraph on the page, because it is the one that tells you what the criteria actually mean.
How we are funded
Affiliate commission. When you buy through a link here we earn a percentage, at no extra cost to you. Amazon is the only network we are in as of 2026-09-01, and the tag is bristleandbot-20. Commission does not affect what we recommend or in what order, and it cannot, because the criteria are published before the products are chosen and the sources are linked so you can check the reasoning. The full statement is on the affiliate disclosure.
Prices, ratings and review counts
Prices render only from a live data feed, and only while that feed is fresh. If it is stale or unavailable, the button reads Check price on Amazon and no number appears. We never print a remembered price, a cached one, or one we saw last week. The same rule applies to star ratings and review counts: where they appear they come from the live layer and are attributed to Amazon, never authored by us.
How to correct us
Email info@bristleandbot.com. If a figure is wrong, or a manufacturer publishes something we have marked as not published, we will change it, add the source, and date the change. Corrections are dated on the page rather than silently edited — the policy is on editorial policy.
Questions people actually ask
+ Do you test the robot vacuums you recommend?
No. We have never owned or handled any of them. We compile published specifications, cite every individual figure, and say plainly where a manufacturer publishes nothing. That is a weaker claim than a lab makes and we would rather make it honestly.
+ Then why should I trust your rankings?
Because you can check them. Every number links to the page it came from, every ranking states its criteria before naming a product, and every roundup names something to skip. If a number is wrong, tell us and we will correct it with a date.
+ Why do some cells say not published?
Because the manufacturer does not publish that figure. We could fill those cells from a review site or an estimate; we would rather show you the gap, since which makers publish what is itself one of the more useful things on this site.
+ Does affiliate commission affect your rankings?
No, and the structure is what stops it: criteria are published before products are chosen, sources are linked so the reasoning is checkable, and every roundup names a product to skip. We are also on a single network, so there is no commission-rate arbitrage between retailers to be tempted by.
+ Where do the prices come from?
A live data feed, rendered only while it is fresh. When it is stale or unavailable the button reads Check price on Amazon and no number is shown. We never print a remembered price.