How CompareSmartDevice Compares Robot Vacuums: Our Methodology
Last updated: Aug 21, 2026 · Category: Robot Vacuums
Our spec-based methodology: manufacturer pages as primary sources, check dates on every claim, fail-closed verification, and no hands-on testing claims.
Affiliate disclosure
As an Amazon Associate I earn from qualifying purchases. At launch, purchase links on this page will be direct, tagged Amazon Special Links. Until our owned Associates account is approved, model and price data come from the Amazon Creators API catalog validated 2026-08-18, and no affiliate tag is attached yet.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on the relevant Amazon Site at the time of purchase will apply to the purchase of this product. (Prices checked 2026-08-18.)
Every comparison on this site is a spec-based comparison: we compare numbers that manufacturers publish, each tied to a source and a check date, instead of claiming hands-on results before we have actually tested. This page explains exactly which sources we use, in what order, how every claim is cited, and what we deliberately do not claim yet.
The method in three sentences
- Specs come from the manufacturer's own product page, fetched directly and recorded with a check date — for example, 18,500 Pa suction for the Roborock Qrevo S Pro (us.roborock.com, checked 2026-08-18).
- Prices come from the Amazon Creators API catalog, not from scraping product pages; every price carries an "as of" date — for example, all launch prices as of 2026-08-18.
- If we have not verified a value, we say so ("not yet verified") instead of guessing. That is the fail-closed rule: uncertainty stays visible rather than becoming a silent fact.
What does "spec-based comparison" mean?
A spec-based comparison ranks and contrasts products using published specifications — suction in pascals, dock functions, mop systems, dimensions, warranty terms — each attributed to the page where we read it. It is not a hands-on review: we have not measured cleaning performance, runtime, or noise for our launch models, so no article on this site implies that we did. We also do not display Amazon star ratings or review counts, because Amazon's participation requirements restrict how affiliate sites may present them.
Which sources do we use, and in what order?
We rank sources by closeness to the original fact. A claim is published at the highest tier that supports it; lower tiers are labeled as such when used.
| Tier | Source type | Used for | Example from our launch set |
|---|---|---|---|
| 1 — Primary | Manufacturer official product page (fetched directly) | Suction, mop system, dock functions, dimensions, stated warranties | 30,000 Pa for the ECOVACS DEEBOT T90 PRO Omni (ecovacs.com, checked 2026-08-18) |
| 2 — Catalog API | Amazon Creators API catalog records | Model existence, ASIN, US marketplace price, availability | $599.00 for the DEEBOT T90 PRO Omni (Creators API, as of 2026-08-18) |
| 3 — Official secondary | Manufacturer help pages and app store listings by the manufacturer | App feature availability, subscription statements | iRobot Select status "not accepting new subscriptions" (select.irobot.com, checked 2026-08-18) |
| 4 — Labeled secondary | Community or press sources quoting the manufacturer | Corroboration only, always labeled as secondary | Roborock support statement on free app functionalities quoted in a community thread (reddit.com/r/Roborock, checked 2026-08-18) |
What do we verify for each model?
The table below lists the fields in our normalized product record and how each one is verified.
| Field | Verification method | Citation format on pages |
|---|---|---|
| Price (US marketplace) | Creators API catalog record from the validation sweep; no scraping of product pages | "$599.00 (Creators API, as of 2026-08-18)" |
| Suction (Pa) | Value extracted from the manufacturer's own product page during a dated fetch | "30,000 Pa (ecovacs.com, checked 2026-08-18)" |
| Mop system | Pad type and wash details extracted from the manufacturer's own product page | "75 °C hot-water mop wash (us.mova.tech, checked 2026-08-18)" |
| Dock / self-empty functions | Self-empty interval, water refill, pad wash/dry extracted from the manufacturer's own product page | "Up to 100 days hands-free (dreametech.com, checked 2026-08-18)" |
| Spec status flag | Verified versus pending per model; pending means the field has not yet been extracted from a primary source | "Not yet verified" in comparison tables |
How are claims dated and cited?
Every specification claim carries two things: the domain where we read it and the date we checked it. Every price mention carries an "as of" date, because marketplace prices move. We use check dates rather than continuous tracking on purpose: Amazon's participation requirements prohibit price-tracking functionality without a separate written agreement, so our launch pages record point-in-time values with explicit dates instead of historical curves.
What do we not claim yet?
- No hands-on testing claims — no model in the launch set has been purchased and tested under a published test protocol.
- No measured runtime or noise figures — where manufacturers state these, we may quote them as manufacturer-stated; otherwise they are pending.
- No star ratings or review counts — excluded to stay inside Amazon's display restrictions for affiliate sites.
- No price history — static "as of" dates until a compliant tracking arrangement exists.
How are "best-for" picks decided?
A pick is published only when it rests on verified data or catalog facts, and every pick states its reason in terms of cited numbers. For example: the Dreame L40 Ultra Gen 2 is our least-base-maintenance pick because Dreame states up to 100 days hands-free (dreametech.com, checked 2026-08-18), the longest verified interval in our set; and the iRobot Roomba 105 Vac at $149.00 is a budget pick on catalog price alone (Creators API, as of 2026-08-18), with its unverified specs flagged rather than smoothed over.
What happens when a spec or price changes?
We re-fetch the source page, update the value and its check date, and record the change in our corrections log. If a previously verified value can no longer be confirmed on the primary page, we fail closed: the field returns to "not yet verified" until it is re-extracted.
Frequently asked questions
Do you test robot vacuums hands-on?
Not in the launch phase. Every comparison is labeled spec-based, meaning we compare manufacturer-published specifications with sources and check dates instead of measured performance.
Where do your prices come from?
From Amazon Creators API catalog records collected during our validation sweep — for the launch set, as of 2026-08-18. We do not scrape product pages, and we do not track price history at launch.
What does "not yet verified" mean in your tables?
It means we have not extracted that value from a primary source as of the check date. It is a deliberate marker for pending data, not an absence — and it prevents us from publishing a guess as if it were a fact.
Sources referenced in this methodology
- Manufacturer product pages for iRobot, Roborock, Shark, ECOVACS, Dreame, eufy, and MOVA — all checked 2026-08-18 (see each article's source list).
- Amazon Creators API catalog validation (US marketplace), as of 2026-08-18.
Published: August 18, 2026. Updated: August 18, 2026 (initial draft).
Where to Check Current Prices
\nProduct prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on the relevant Amazon Site at the time of purchase will apply to the purchase of this product.
\niRobot Roomba Plus 405 (G181) Combo robot + AutoWash dock
\n- 70X more intense suction (vs. Roomba 600 series)
- ClearView LiDAR
Roborock Qrevo S Pro (2026)
\n- 18500 Pa Pa
- PreciSense LiDAR (LDS) + Reactive Tech structured-light obstacle avoidance
ECOVACS DEEBOT T90 PRO Omni
\n- 30000 Pa (16 L/s airflow)
- 140 min silent / 120 min standard (vacuum only, hard floor); 350 min sweep & mop standard with Perpetual Run
- Mini-ToF navigation + AIVI 3D 4.0 AI obstacle avoidance (structured light, RGB)
Dreame L40 Ultra Gen 2 Robot Vacuum
\n- 25000 Pa Pa
- up to 231 min (Vacuum Only mode, Quiet Suction)
- LDS navigation + 3D structured-light obstacle avoidance
As an Amazon Associate we earn from qualifying purchases.
More in Robot Vacuums
Research methodology
This article was generated from verified product data using our research pipeline. Comparisons cover 12 weighted criteria, with data completeness and research confidence reported separately from performance. Claims are cross-referenced against manufacturer specifications and Amazon API data.