How We Test

Every figure in a review comes directly from our own hands-on testing of the machine in real US homes - never from a press release, a sponsor, or a forum thread. We treat three primary sources as authoritative:

• Our bench measurements - suction in Pa and air watts, runtime on a stopwatch, weight on a scale, and noise on a meter. We record each reading and note the day we took it.
• Our real-floor test runs - the same machine pushed across hardwood, carpet, and pet hair, so we capture how it actually picks up rather than how the box claims it does.
• Filter and brushroll teardowns - we open the machine, pull the filters, and check what they actually trap, plus how the brushroll handles hair and debris. We photograph the teardown and note the model and date.

Suction & Power

• Suction: the sealed suction we measure in Pa and air watts, and how it holds up as the bin fills and the battery drains.
• Power modes: how eco, auto, and boost modes change pickup and how long each one lasts in practice.

Runtime & Handling

• Runtime: the minutes we time on a stopwatch in each power mode, and how that compares to the battery rating on the box.
• Weight & maneuverability: what the machine weighs on our scale, how it feels overhead and under furniture, and how it turns on carpet.

Filtration & Noise

• Filtration: the filter stack and whether it is true sealed HEPA, plus what the teardown shows it actually traps.
• Noise: the level we read on a meter in each mode, and whether it is a pitch you can live with day to day.

By Surface

• How it picks up: hardwood, low-pile carpet, and embedded pet hair - and whether the brushroll tangles, scatters debris, or self-empties cleanly into the dock.

When a brand launches a new model, we re-test the affected machine and re-baseline any ongoing comparisons that reference it.

A review with no date is useless - a suction number only matters if the date is current. So every vacuum guide we publish carries a visible “Last updated” date showing the day we last tested the machine by hand.

What the Date Means

The “Last updated” date is not the publish date and not an automatic timestamp. It is the day a human on our team last put the machine through our tests and confirmed that every figure in our guide - suction, runtime, filtration, by-surface pickup - still matched. If a detail changed, the guide was updated before the date was moved.

Why It Matters

Brands quietly revise brushrolls, push firmware that changes runtime, and replace a model with a near-identical successor without making noise about it. The date is how you tell whether what you're reading reflects the machine as it ships today or last year. We would rather show an honest, slightly older date than fake recency.

While the specifics vary by machine, every vacuum follows a consistent core process:

1. 1Unbox and bench it - we set up the machine, measure its suction in Pa and air watts, weigh it, and read its noise on a meter.
2. 2Run it across real floors - we push it over hardwood, carpet, and embedded pet hair, noting how it picks up, scatters, or tangles in each power mode.
3. 3Time the runtime - we run the battery down on a stopwatch in each mode and check the result against the rating on the box, not assumed.
4. 4Tear down the filters - where a filtration claim is contested or surprising, we open the machine and check what the filters actually trap so readers can see the evidence.
5. 5Date-stamp it- we record the test date and attach the “Last updated” stamp shown on the published guide.
6. 6Log the details - the suction, runtime, filtration, weight, and any test notes are recorded in a standardized internal sheet that the fact-checker and senior editor reference during the editorial review.
7. 7Editorial sign-off - a second editor confirms every figure traces back to our own test data before the guide is published.

Testing isn't a one-time event - lineups change, so we re-check machines on a cadence and whenever something prompts it.

• Scheduled re-testing - major brand and model hubs (Shark, Dyson, Roomba, Roborock, and the like) are re-tested by hand on a regular cadence, and the “Last updated” date is refreshed each time.
• Trigger-based re-testing - when a brand launches a new model or ships a firmware update, or a reader reports that the machine no longer behaves the way our guide describes, we re-test the affected guide right away.
• Seasonal re-testing - buying guides and best-of roundups are re-tested ahead of the big sale seasons, since new models land and prices shift every year.

Transparency means being honest about the limits of testing, not just the strengths.

• Marketing claims- we don't present a brand's headline suction or runtime number as fact. If our own measurement doesn't match the box, we say so.
• Stale specs- we don't publish a suction or runtime figure without a “Last updated” date. A figure with no date is treated as untested.
• Crowdsourced spec data - we do not average forum claims, user reports, or comment-section anecdotes into a published review. Those may flag something worth re-testing, but the figures themselves always trace to our own hands-on testing.
• Guaranteed results - testing a vacuum tells you how it performed in our homes, not that it will behave identically in yours. Where results vary by floor or setup, we say so and point you to our troubleshooting and repair guides.