Methodology

WhichVac does not run dirt tests. Every 0–100 score is arithmetic on specs a cited page published. If a number is missing, the page says Not published and that input drops out of the average. It is not filled in.

What a score is

• Cleaning power mixes suction, mop hardware, navigation, and obstacle sensing.
• Smart features mix navigation, obstacle sensing, and app features a source confirmed.
• Low maintenance mixes the self-empty dock, bin interval, mop wash, mop dry, water refill, a described hair system, and carpet lift.
• Value is that performance blend divided by the lowest recorded price, scaled so the best raw value in this catalog is 100.

Weights

Cleaning

Suction 0.42, mop 0.28, navigation 0.18, obstacle avoidance 0.12. Suction is linear to 25,000 Pa, then capped.

Mop points

None 18, pad 50, sonic 66, dual spinning 84, roller 94. A confirmed combo with an unnamed mop uses 55 so the score does not rise just because the mechanism is missing.

Navigation points

Bounce 32, gyro 48, camera 70, LiDAR 84, LiDAR plus camera 96.

Obstacle points

Bump 28, infrared 48, structured light 74, camera 88, structured light plus camera 96.

Smart features

Navigation 0.35, obstacle sensing 0.35, confirmed app features 0.3. Each confirmed feature among mapping, no-go zones, room selection, schedules, voice assistants, multi-floor maps, and dirt detection adds 16 points, capped at 100. Features nobody mentioned are left out. They are not scored as absent.

Low maintenance

Weights: self-empty 30, dock interval 15, mop wash 18, mop dry 14, refill 10, hair system 8, carpet lift 5. Vacuum-only models skip the mop chores instead of being punished for them. Dock interval reaches 100 at 90 days. A cited “no self-empty dock” scores the interval at 25, and that stand-in is labeled an assumption.

Value

Performance = 0.5 × cleaning + 0.25 × smart + 0.25 × maintenance. Raw value = performance per $100. The highest raw value in the catalog becomes 100. A new price can reshuffle the list. Until a live feed exists, every value score is marked as using a snapshot or list price.

Pet hair and apartments

Pet hair weights suction 0.5, self-empty 0.2, a described hair system 0.2, carpet lift 0.1. Apartment fit weights height 0.5, dock bulk 0.35, runtime 0.15. A self-empty dock with no published height scores 40 for bulk. That is an assumption, printed on the breakdown, not a measurement.

Lists

Numbered lists keep models whose relevant score has at least 60% of its weight filled by published inputs. The rest are listed underneath. Best value also keeps at most two models per brand in the top 10. Guide score 100 means first on that list. It is not a second rating. Under $300 uses the lowest recorded price, and a live price can move a model on or off the page. Obstacle avoidance omits models whose sensor class was not named.

Prices

Seed prices are cited snapshots from the pages named on each spec, or they are blank. None of them are shown as Amazon’s price. List prices are kept separate and are used for “from $X” only when no street snapshot exists. The Amazon row is filled only by a live quote from the Creators API, which runs first, or from PA-API 5 when Creators has no quote for that model. Sample mode, the default without those keys, does not invent a number.

Worked example

Roborock Q7 M5+ has no published pascal figure, so suction is left out and the cleaning score is marked partial. Coverage 28%.

Sources are linked on each spec row. WhichVac did not retest these robots. A wrong citation should be reported to the contact address in the footer.