Methodology

Sources

Every figure starts on an official page: the maker's product page, manual or specification sheet. Each model lists those pages and the date we read them. Retail sites help us find models. We take a fact from a retailer in one case only: a list price for a model the maker sells through retailers alone, marked as a retailer price. The data page shows how many records publish each field.

The four verdicts

• Fits: every check passes with room to spare.
• Tight: nothing rules the model out, but one check passes with less than the margin.
• Needs a check: nothing rules the model out, but the maker does not publish a value a check needs. We never pass or fail a model on a value we do not have.
• Does not fit: at least one check fails. The line under the model gives the reason in numbers.

A model goes to does not fit when any check fails, to needs a check when none fails but one is unknown, to tight when none fails or is unknown but one is at its limit, and to fits otherwise. Notes add information and never change a verdict.

Robot mower sizer

The sizer only considers models sold in the region you choose in the header. Where a maker rates or names a machine differently by market, the sizer uses the figure for your region; where its figures conflict, the dataset keeps the lower one.

Area

A rating does not say how much beds, corners and short mowing hours cost. Real lawns have beds, trees, narrow strips and hours when the mower may not run, so the sizer multiplies your lawn area by two factors to get the rating it needs.

Rating needed = lawn area × shape factor × hours factor. An 800 m² (8,611 sq ft) lawn of typical shape, mowed at any time, needs 800 × 1.35 × 1.0 = 1,080 m² (11,625 sq ft). Half an acre (2,023.43 m²) of complex shape, mowed in daytime only, needs 2,023.43 × 1.5 × 1.5 = 4,552.7 m².

Makers state area on different bases, so the sizer compares your lawn with a limit. For a maximum, a per day figure or a figure without a stated basis, the limit is the rating. Kress rates area per day: the lawn a mower cuts in 24 hours. The sizer compares a per day rating as published, with no discount and no factor, and the site prints "per day" next to every such rating. For a recommended area (Segway Navimow, Roborock and ANTHBOT publish these) the limit is 1.2 times the rating, because a recommended size is not a ceiling. The 1.2 allowance is a rule of thumb. The area ratings guide explains the bases.

• Fits when the limit is at or above the rating needed.
• Tight when the limit covers the lawn area but not the rating needed.
• Too small when the limit is below the lawn area. The model does not fit.
• A limit more than 6 times the lawn area gets a note that the mower is oversized. That note never rules a model out.
• Model pages say which lawns a model is sized for: from the limit divided by 1.5 (a complex lawn) to the limit divided by 1.2 (an open one).

Slope

Slope is the steepest part of the lawn the mower has to climb, in percent (rise over run × 100) with degrees alongside. A model passes when its published maximum slope is at least 5 percentage points above yours. It is at its limit when the maximum is at or above your slope but less than 5 points above it, and it fails below your slope. On a 30% lawn, a rating of 35% passes, 32% is at its limit and 27% fails. A model without a published slope limit goes to needs a check.

Other mower checks

• Boundary wire: with "no wire", a model that needs a boundary wire fails. With "either", nothing changes. With "wire is fine and preferred", nothing fails and wire models come first within each group.
• Sky view: a model that steers by satellite positioning alone, with no camera, no LiDAR and no boundary wire, gets the warning "satellite positioning needs open sky" when you choose heavy cover, and a softer note when you choose some trees or buildings. A model that steers by satellite with a camera or LiDAR to help gets no warning, but under heavy cover a note that satellite positioning works best under open sky and that the maker's guidance for tree cover is worth reading. Models that steer by wire, camera or LiDAR alone, and GARDENA's satellite system without RTK (listed as its own navigation system), get no sky note. Sky never rules a model out.
• Narrowest passage: a note, never a failure. When the maker publishes the narrowest gap the mower can pass and it is wider than yours, the note warns with both numbers; when it is narrower, the note confirms it; when it is not published, the note says so. Only 17 of 65 mower records publish this width, too few to rule models out on it.
• Separate lawns: with more than one, a model fails when its maker says it handles a single area only, and goes to needs a check when the maker does not say. A model that handles several areas gets a warning when the maker states a lower number of areas than yours, and a note when it gives no number.
• Budget: a model fails when its price in your region is above your budget. A model without a price in your region gets the note "price not published for this region" and is never ruled out for it.

Mower ranking

• Fits, tight and needs a check: the limit closest to the rating needed comes first, so inside fits the smallest limit that still fits leads. Ties go to the lower price in your region (models without one last), then the larger slope margin, then the name.
• Other orders you can choose: lowest price, most headroom (the largest limit compared with the rating needed) or steepest slope rating. Ties follow the closest fit.
• Does not fit: the fewest failures first, then the smallest shortfall in area.

Robot vacuum finder

The finder only considers models sold in your region. Lengths are compared in millimeters.

• Threshold: passes when the published climb height is at least 2 mm (0.08 in) above your tallest threshold, is at its limit when it is at or above the threshold but less than 2 mm above, and fails below it. With a 20 mm threshold, a 22 mm climb passes, 21 mm is at its limit and 18 mm fails. A threshold at its limit makes the model tight. A model without a published climb height goes to needs a check.
• Furniture: when you give the gap under your lowest furniture, a model passes when its height plus 5 mm (0.20 in) is at or below the gap, and fails otherwise. Under a 95 mm gap, a 90 mm robot passes and a 92 mm robot fails. There is no limit band for height. A model without a published height goes to needs a check.
• Mopping: when you want mopping, a vacuum without a mop fails. On mixed floors or carpet the finder notes how far the mop lifts, warns when the lift is not published and the mop does not detach on its own, and warns on thick pile unless the mop detaches on its own. On hard floors there are no mop notes. When you do not want mopping, nothing fails and models that mop still appear.
• Dock: docks rank as charging only, self emptying (empties the bin) and full service (empties the bin, washes and dries the mop). A model fails when its dock does less than you ask for, and a robot without a dock fails any dock request.
• Dock space: a warning, never a failure, when a published dock width or height is larger than the space you give, and a note when the dimension is not published. Only 10 of 42 vacuum records publish a dock width, and heights rest on our reading of unlabeled dimensions.
• Floor area: optional. When you give it, a note says whether one charge covers it, whether one charge falls short, whether the robot finishes after a recharge, or that coverage is not published. Coverage never rules a model out.
• Levels: a note when the robot stores fewer maps than you have levels.
• Pets: a note suggesting a self emptying dock when the dock does not empty the bin.
• Budget: as for mowers. A missing price is a note, never a failure.

Suction in pascals appears in the specifications and never in a verdict or a ranking: makers measure suction in different ways, so the numbers do not compare. Navigation is grouped as LiDAR turret or no LiDAR turret, because makers label the other designs in different ways.

Vacuum ranking

• Fits, tight and needs a check: the lowest price in your region first, models without a price for your region after the priced ones, then the larger threshold margin, then the lower robot, then the name.
• Does not fit: the fewest failures first, then the same order.

Ranking and money

The rules above set the order. Whether a link can earn us money never changes the order, the wording or which models appear. See how we make money.

Pages built from the same rules

Scenario pages, such as a half acre lawn or a home with thick carpet, run fixed inputs through the same engine when the site is built and show the inputs, with a link that opens the tool with them. A scenario page exists only when at least three models fit it in its default region. Brand pages and their tables of which model suits which lawn or home use the tools' defaults on standard lawn sizes and thresholds. Every one of these pages ranks the models in the Botsize dataset and names how many it ranks.

What we do not know

• Specifications describe a model, not your garden or your home. Grass type, soil, wet slopes, satellite signal in your yard and rug edges all change how a robot behaves, and makers do not publish figures for them.
• The shape and hours factors, the slope and threshold margins, the furniture gap and the allowance for recommended ratings are rules of thumb. No maker publishes how its rating shrinks on a complex lawn.
• Makers rate area on different bases and measure suction in different ways. We state the basis next to each area rating and keep suction out of every verdict.
• A value a maker does not publish shows as not published, and a check that needs it sends the model to needs a check. A field that is empty for most models appears in the specifications only and never filters or ranks a result.

Prices

Prices are the maker's list price in each market, or the price its page showed that month (sometimes a sale price), in that market's currency, with the month we read it. Each price says which of the two it is. We never convert currencies. Prices change often, so confirm the price on the maker's page before you buy.

Corrections

To report a wrong number, send the maker's page that shows the right one to hello@botsize.com. We check every report against that page.