Editorial Policy

This page documents how ApplianceWatchDog content gets researched, checked, published, corrected, and kept current. It applies to every page on the site.

• Primary sources first. Coverage facts come from provider contracts and official pages; complaint data from BBB files; review scores from the platforms themselves; regulatory facts from regulator records (e.g., state attorney general press releases and filings); ownership facts from company and SEC filings; cost benchmarks from named published datasets (HomeGuide, Angi, HomeAdvisor, This Old House, NAHB).
• Everything is access-dated. Volatile figures — prices, fees, caps, scores, complaint counts — carry an "as of" stamp on-page (currently August 2026) and are re-verified on the cadence below.
• Customer quotes are paraphrased and labeled with platform and approximate date. We don't invent quotes, composite them without saying so, or present marketing claims as customer experience.
• What we won't use as a source of record: brand press kits, affiliate-network sales materials, or any statistic a company can pay to change.
• Estimates are labeled. Where a figure is derived (e.g., a metro cost modifier applied to a national range) rather than directly published, the page says "estimate" and shows the inputs.

Every page is checked before publication by someone other than its writer: numbers against primary sources, claims against contracts, safety steps against our technician reviewer's trade experience (team). Pages that state legal or regulatory facts get a second read against the underlying documents. A review doesn't publish until its score reconciles exactly with the published methodology.

• Report an error: /contact, subject "Correction: [page URL]". Anyone can — readers and rated companies alike.
• Our commitment: we verify against primary sources and correct confirmed errors promptly — target one business day for material figures.
• Transparency: material corrections (anything that could have affected a reader's decision — a wrong price, cap, score, or legal fact) are noted on the corrected page with the date and what changed. Typos and formatting fixes are made silently.
• Score disputes are different from factual errors. We fix facts on evidence; scores change only when underlying data changes at re-review. No one can negotiate a score — see how we make money.

We use AI tools in our editorial process — for research organization, drafting assistance, and consistency checks across our data tables. Every published page is directed, reviewed, fact-checked, and approved by human editors, who are accountable for its accuracy; every factual claim is verified by a person against the sources above, and ratings are set by the editorial team through the published methodology, never generated automatically. AI is a drafting tool here, not an author of record and not a fact-checker.

• We read the actual plan contracts — caps, exclusions, fee schedules — not just the marketing pages.
• We read complaint files and review distributions on multiple platforms, and we disclose when a platform hosts paid placements for the brand being reviewed.
• We track regulatory dockets; enforcement actions trigger immediate review updates under the methodology's hard rule.
• We disclose what we haven't verified: figures we couldn't independently confirm this cycle are flagged on-page ("verify at refresh") or omitted.
• We accept no sponsored reviews, no pre-publication approval by brands, and no compensation contingent on content.

Quarterly re-verification of all scores and volatile figures; out-of-cycle updates for regulatory actions, ownership changes, major price moves, and confirmed corrections. Every page displays its last-updated date.

Commercial relationships are managed separately from editorial work; writers and editors score without access to commission data. The full policy — and the receipts — live at /legal/affiliate-disclosure.