Methodology

Provider inclusion

A provider is included when a dated audit finds a live first-party site, an explicit Utah service area, at least one covered exterior-cleaning service, and a sufficiently clear business identity. Generic lead directories and unresolvable duplicates are excluded. The audit is not a guarantee of completeness.

Evidence labels

Every provider is checked for a Utah state entity or local official business record, operating history, insurance statement, Google rating, BBB profile, provider-supplied before-and-after gallery, published pricing, service-area evidence, and method or runoff detail. A local licence or directory entry is not presented as a state entity match. Provider-stated facts remain attributed. Official data is identified by geography and date. Third-party snapshots name the publisher and audit date. Not publicly specified means the reviewed sources did not answer the question; it does not mean the provider lacks the capability.

Professional news desk

Legal, environmental, safety, and recall claims begin with the responsible government authority or controlling document. Utah coverage identifies the state or local layer checked because a municipality, sewer authority, water supplier, or site permit can change the answer for a specific job.

Association, manufacturer, and trade reporting is labeled as such. Short source updates use an original summary and send the reader to the publisher; source article bodies are not copied. First-party field guides state publication and review dates, list their sources, preserve proposal-versus-final status, and separate practical editorial inferences from source claims.

Ratings

Google, Yelp, and BBB values are treated as independent snapshots only after a confident name-plus-contact, website, or location match. A record includes its source URL and access date. A rating repeated on a provider’s own site is labeled provider-stated until the live platform listing is independently captured. BBB letter ratings and accreditation are kept separate from customer-review averages. No review text is copied, and no rating affects the default alphabetical order.

Local data

Legal place boundaries come from UGRC and Census sources. Weather context uses the nearest NASA POWER 2001–2020 regional climatology grid point and is labeled as planning context, not a station, parcel, or live forecast. ACS facts appear only when collected from their named geography. Water hardness is not inferred from a city name: users are directed to the serving utility report or a property test, and TDS is never relabeled as hardness.

Cost model

Residential and commercial models are maintained separately. Each released benchmark cell is keyed by service, project type, and a coarse Utah metropolitan area. When a cell is too sparse, the estimator follows a visible fallback chain to a broader Utah cell and then to the national or editorial planning baseline; it does not silently invent a local multiplier.

Published provider prices, voluntarily submitted completed-job records, and contractor survey answers remain separate evidence channels. A provider's advertised minimum or asking rate is not relabeled as a completed-job price. Completed-job records must describe one service, a final paid total, a compatible quantity, a completion month, and controlled inclusion flags. Contractor surveys ask about historical completed work, not coordinated future pricing.

Raw submissions enter a pending review queue and never change the live estimator. Review rejects exact or suspected duplicates, incompatible units, bundled totals without a service allocation, implausible values, and records that cannot be interpreted consistently. Public metro cells require at least 10 qualifying records in one primary completed-job or contractor-survey channel, adequate source diversity, a documented inclusion rule, and repository approval; provider statements remain a comparison stratum and never count as completed jobs. Released snapshots report channel-specific strata, their evidence mix, collection window, sample size, and fallback status. A non-primary channel statistic remains suppressed below five observations even if the primary channel qualifies.

Saved estimates are prospective private planning records. They are not mined for benchmark research, and reaching or saving an estimate never submits a completed-job record. Benchmarks, minimums, model versions, and multipliers are visible; editorial assumptions remain explicitly marked. The estimator is never a contractor bid.

Corrections and refreshes

Provider discovery, local-data enrichment, rating snapshots, source updates, and news articles are reviewed repository changes. The public site does not run an unattended news scraper. Material edits update the review date and validation suite.