Our Review Methodology
This page explains, in specific terms, how a product ends up on Robot Vacuum Specs and how we form the opinions in each review — including what we do not do, so you can judge how much weight to give our recommendations.
What this methodology is (and is not)
Robot Vacuum Specs evaluates products based on the manufacturer's published specifications and product documentation, cross-checked against the live Amazon listing at the time of publishing. We do not currently conduct in-house physical or laboratory testing of the robots we cover. We deliberately avoid language like "we tested" or "hands-on" anywhere on this site unless that specific evidence actually exists for that specific product — most of our review copy instead uses phrasing like "rated," "the manufacturer states," or "based on the listed specifications" for exactly this reason.
Firsthand testing
We do not currently operate a dedicated testing program. When a product is available for firsthand evaluation, any direct observations will be identified as such, clearly separated from manufacturer specifications and editorial interpretation. Until then, we do not present specifications or editorial analysis as physical test results.
The categories we evaluate
| Category | What it means | Evidence we use |
|---|---|---|
| Cleaning | Suction power and brush design as they relate to debris pickup | Manufacturer-rated suction (Pa) and brush/roller design described in the listing |
| Navigation | How the robot maps and moves through a home | Stated navigation technology — LiDAR, camera-based/vSLAM, or none — and any independent confirmation of how that technology generally behaves (see our navigation technology guide) |
| Obstacle avoidance | How the robot handles cords, pet waste, shoes, and clutter | Manufacturer claims about object-recognition capability and the number/type of objects it is rated to detect |
| Mopping | Whether and how well the robot handles hard-floor mopping | Mop mechanism (pad type, spinning vs. static), auto-lift behavior on carpet, and water tank capacity as stated by the manufacturer |
| Battery | Runtime and charging behavior | Manufacturer-stated runtime and charging specifications — we do not independently measure real-world runtime |
| Noise | How loud the robot is in normal operation | Manufacturer-stated noise levels where published; otherwise we do not make a specific noise claim rather than guess |
| App experience | What the companion app lets you actually do | Documented app features — scheduling, room selection, no-go zones — from the manufacturer's own product and app documentation |
| Maintenance | What owning the robot actually requires day-to-day | Stated self-empty/self-wash capability, filter and consumable-part design, and documented maintenance intervals |
| Features | Smart-home integration, multi-floor mapping, and similar extras | Manufacturer-documented compatibility (e.g. Alexa) and feature lists |
| Value | Whether the feature set justifies the price tier | Editorial comparison against other currently-reviewed products in a similar price tier — see our budget vs. flagship comparison for this in practice |
How we form comparisons and rankings
When we compare two or more products (in a roundup, a head-to-head comparison, or the "alternatives" section of a review), we compare the same categories above using each product's own published specifications side by side. Where a manufacturer does not publish a figure for a given category — for example, a suction rating that is not listed — we say so explicitly rather than substitute an estimate.
How we handle missing or ambiguous information
If a listing is ambiguous about whether a feature is fully automatic versus something the manufacturer describes more vaguely, we describe it in the more conservative terms the listing actually supports, and note the ambiguity rather than assume the more flattering interpretation. We do not fabricate a specification, statistic, or feature that is not present in the manufacturer's own materials.
Where the numbers in our databases come from
The specification database, its focused tables (suction, thresholds and clearance, dock automation) and the cost calculator all read from a single data file, so a figure never differs between a review, a comparison, and a database page. Every value is a manufacturer-stated specification, taken from the manufacturer's product page or the Amazon US listing text and re-checked by a person on the date shown on each record.
Three conventions keep this honest. "Not stated" is not "no": it means the source did not publish the figure. A rated figure is not a measurement: for example, rated suction in pascals is a manufacturer claim with no shared test method across brands, so we never treat it as a cleaning score. No prices, ratings or availability are stored — those change constantly and are shown only on the live Amazon listing, or entered by you in the calculator.
How we word comparisons and recommendations
Every comparison labels what kind of statement it makes: a manufacturer specification (a figure the maker states), a documented feature (something the maker says the product does), an editorial interpretation (our reading of what that means for a type of buyer), a calculated value (arithmetic on stated figures, with the inputs shown), or a hands-on measurement. We perform no hands-on measurements, so that last label is always "not performed".
Where a page says a robot is a "better documented fit" for a use case, it names the documented feature behind that and does not claim the robot cleans better. We do not use prices or Amazon ratings to rank products, because both change too often to keep accurate.
Corrections and updates
If you find an error, tell us through the contact page with the page address and the source you are relying on. We check it against the manufacturer's documentation and correct the page and the underlying data record together, so a fix applies everywhere the figure appears. Pages show the date they were last updated, and each specification record carries the month it was last reviewed. We change a page's date only when its content actually changed.
What updates when
We generally do not display product prices because Amazon prices and availability change frequently. Each product page links directly to Amazon so readers can check the current price and availability — see our affiliate disclosure for how that works. Editorial content (verdicts, comparisons, guides) is updated when a meaningful change occurs — a product is discontinued, a specification changes, or we identify an error — not on a fixed schedule for its own sake.