How we judge — and what we refuse to claim

The rubric

Six criteria, weighted before any model is scored. Weights reflect what owners actually live with after the first week: the wash itself dominates, because a bidet that washes poorly is furniture; hygiene follows, because the nozzle lives where it lives; and total cost outranks controls, because a confusing remote is an annoyance while a dead seat with no parts is a loss.

What each criterion actually examines

Wash performance (22%)

Stream architecture and adjustability, as documented: pressure adjustment range, nozzle position steps, oscillation and pulsation modes, aeration, and dedicated wash modes (rear, front, vortex/enema-style where offered). A model earns points for range and control, not for the number of marketing names given to the same spray.

Hygiene & self-cleaning (18%)

Nozzle material (stainless steel scores above plastic), self-cleaning cycles, sterilization systems with a documented mechanism (electrolyzed water, UV, silver-ion), nozzle retraction and guarding, bowl premist systems, and whether the seat quick-releases so the area around the nozzle can actually be cleaned by a human.

Comfort & heating (18%)

Heating architecture and its consequences: tank reservoirs are warm instantly but finite; tankless instant heaters are endless but draw heavy current at the moment of use; hybrid systems split the difference. Also: seat heating range, warm-air drying, deodorizer type, night light, and lid/seat closing behavior.

Build, warranty & support (16%)

Warranty length and coverage shape — a three-year full warranty outscores a three-year tiered one, and both outscore one year. Parts availability (replacement nozzles, filters, remotes), electrical and plumbing certifications (UL/ETL listing, IAPMO cUPC), and the manufacturer's time in the category.

Total cost of ownership (14%)

Five-year cost, not sticker price: street price, installation, electrical work where required, and consumables — deodorizer cartridges, water filters where fitted, nozzle tips. A model is scored within its class, against what its class should cost; the class is always printed next to the score.

Controls & usability (12%)

Remote vs. side-arm vs. dial, judged on learnability: can a guest use it without instruction? User presets, control legibility, and accessibility considerations — raised tactile controls and simple single-dial operation score well here, which is why a well-made attachment can beat a feature-dense seat on this criterion.

Scoring anchors

Anchors keep scores honest across models and across time:

• 9–10 — Class-defining. The criterion against which others are measured.
• 7–8.9 — Strong. Minor compromises, none disqualifying.
• 5–6.9 — Adequate, with real compromises a buyer should know by name.
• Below 5 — The criterion fails in a way that affects daily use.

Verdict bands: 8.0 and above — Recommended. 6.5 to 7.9 — Worth considering, for the buyer the scorecard names. Below 6.5 — Not recommended at the checked price; a deep discount can reopen the case, and the scorecard must say at what price.

Evidence admitted

Three classes, in descending order of authority:

1. Class A — Manufacturer documents. Spec sheets, installation manuals, and warranty texts published by the manufacturer. The basis for every feature claim.
2. Class B — Certification listings. IAPMO, UL, ETL and equivalent directories. The basis for safety and compliance claims.
3. Class C — Retailer listings. Used for street prices only, always dated, always expressed as a range when listings disagree.

Review-aggregator star ratings, influencer content, and manufacturer marketing copy are not evidence and are never cited as findings. Where a widely repeated claim cannot be traced to Class A or B evidence, the verdict either omits it or prints it labeled as unverified.

The honest limits of a researched verdict

A spec sheet cannot tell you how a spray feels, how loud a dryer is at 2 a.m., or whether a seat's plastic yellows in year three. Researched verdicts are strongest exactly where buying decisions are most often botched — fit, features, heating architecture, warranty shape, and total cost — and weakest on long-term durability, which only time and ownership data establish. We score what the evidence supports, flag what it cannot, and never borrow the language of hands-on testing (“in our tests”, “we found”) for desk research. When this desk begins publishing ownership-data follow-ups, they will be labeled as a separate evidence class, with their method stated.

Corrections & updates

Verdicts are dated. Price checks are re-run and re-dated; model revisions and discontinuations are noted on the scorecard with the date of the change. Factual errors, when found, are corrected in place with a note — the record matters more than the embarrassment.