How We Rate

We do

• Route buyers through a category path (robot / stick / washer / combo) before model lists
• Score wet-clean class as a hard filter (none / mop pad / wet-dry washer)
• Prefer battery energy in Wh over mAh when both appear
• Archive manufacturer suction claims without cross-brand ranking
• Require a home-test ledger id before publishing numeric home-test fields
• Peer-review technical claims with Marcus Voss (flat fee, no affiliate payout)

We don’t

• Invent home-test stopwatch / pass-rate numbers for unlogged SKUs
• Treat Pa / AW / CFM as a shared industry test standard
• Fabricate customer review quotes (full review text is unavailable from our feeds)
• Sell #1 placement or score by commission rate
• Downgrade an entire editorial page to “specs only” solely because a ledger row is pending — we just refuse fake test numbers

1. DustFree home test — only with home_test_ledger_id
2. Manufacturer spec — nameplate Wh, tank sizes, stated nav class
3. Live retail listing — PartnerBoost / Affscrape verify ASIN + offer
4. Editorial analysis — scene fit, path advice, “who should skip”

HPR (Handpicked Rating) is our 0–10 editorial score on Best-of pages. Default weights for floor care:

• Job / wet-clean fit (30%) — correct class for the brief (washer vs mop pad vs dry-only); hard fails when class mismatches
• Scene ergonomics (25%) — pets / anti-tangle, dock space, multi-floor + stick backup need, noise tolerance for apartments
• Capacity & runtime honesty (20%) — Wh when published, bin / tank sizes, auto-empty / self-clean when the scene needs them
• Street value (15%) — typical US street price vs peers in the same class (as-of dated)
• Listing clarity (10%) — resolvable retail page + coherent published specs. Affiliate payout is not an input.

Each factor is scored 0–10, then combined: HPR = 0.30·J + 0.25·E + 0.20·C + 0.15·V + 0.10·L. Stars on the page are round(HPR / 2) filled stars out of five. Mid-pack units should land near 7–8; 9+ is reserved for clear scene winners.

Worked example (pet apartment, hard floors, light mop)

Suppose a LiDAR robot with mop pad, auto-empty dock that fits a closet, published Wh, and a live US listing at a mid-pack street price:

• J (job fit) = 8.5 — robot + light mop matches; not a greasy-kitchen washer brief
• E (ergonomics) = 8.0 — dock OK, anti-tangle claim present, apartment-aware
• C (capacity honesty) = 8.0 — Wh + bin published; runtime labeled as claimed
• V (street value) = 7.5 — competitive vs peers at write time
• L (listing clarity) = 9.0 — live ASIN + coherent details

HPR ≈ 0.30·8.5 + 0.25·8.0 + 0.20·8.0 + 0.15·7.5 + 0.10·9.0 = 8.2. A washer forced into this dry-apartment brief would fail job-fit hard even with strong marketing suction numbers.

Desk: Ethan Hale. Technical peer review: Marcus Voss. Corrections: contact@dustfreetested.com.

Before Best-of lists, use the public tools — formulas are printed on each tool page (and regression cases live in the repo as docs/tool-tests.md).

• Vacuum type quiz — category path (asset A)
• Scene matcher — relative robot / stick / washer priorities after the path (asset B1)

Scene guides re-weight the same evidence (pets, carpet share, wet class) — they do not invent a second suction league table.

Editorial depth follows US PartnerBoost Joined merchants we can verify (currently strong on Roborock, Dreame, Lefant, YEEDI, eufy and related stick / washer lines). Shark, Dyson, iRobot, and Tineco are not yet Joined under our test media — we will not fake PB rows; coverage expands when catalog access or Affscrape+Amazon fallback is approved.

• Street prices and offers: re-check when rankings are edited; show as-of dates
• Home-test ledger: grows over time; absence ≠ automatic page kill
• Near-duplicate ASINs: primary resolution in data/qa/asin-primary-resolution.md (twins retired; trims tagged)
• Amazon Creators API: currently unavailable