Editorial standards

Every pricing claim, decision framework, and contractor-tactic disclosure on CostPatch follows the standards below. We publish them here so readers can hold us accountable.

1. Data sources (hierarchy of trust)

Pricing figures derive from four source tiers, in priority order:

1. Bureau of Labor Statistics Regional Price Parity (BLS RPP, 2022 series, all-items). Used to derive state-level cost indices from national medians. Public dataset, freely available at bls.gov/regional.
2. Lead-gen platform aggregates from Networx, Bark, Modernize, HomeAdvisor/Angi. Where we have data-sharing agreements or published median-quote data, we anchor our national baseline to these multi-thousand-quote aggregates.
3. Direct contractor surveys — 5–10 phone interviews per service per quarter, asked: "what's a typical price for [X] job in [region] this quarter?"
4. Industry trade publications — Remodeling Magazine, JLC Online, Construction Specifier, Roofing Contractor Magazine. Used for material cost trend analysis (e.g. lumber, asphalt-binder, copper).

When sources disagree (which happens), we publish the BLS-derived range as the headline figure, then note material divergence in the article body.

2. Refresh cycle

Cost data is refreshed on a quarterly cadence. Each article displays a "Last updated" date in the methodology footer. Refresh triggers:

• BLS RPP releases (annual update, typically June)
• Material commodity moves >5% in a quarter (lumber, asphalt, copper, steel)
• New federal tax credits or state subsidies (e.g. IRA energy credits)
• Reader correction submissions exceeding a threshold (≥3 substantive corrections per quarter trigger a full-article review)

3. Fact-checking process

Before publication, every article passes through:

1. Source verification — every numeric claim has a citation, either inline or in the methodology footer. AI-generated draft text is verified against primary sources before publication.
2. Mathematical sanity check — calculator formulas are tested against three known reference jobs (typical, low-cost, high-cost) to confirm output realism.
3. Tone audit — articles are checked for common AI patterns ("delve into", "it's worth noting", "crucial", em-dash listicle structure) and rewritten when found.
4. Tradesperson review (rotating) — once per quarter we send 2 random articles to a working contractor in the relevant trade for blind review. Material corrections are applied within 14 days.

4. AI involvement disclosure

We use AI (Claude by Anthropic) significantly in our content workflow. Specifically:

• Research synthesis — assembling source data into article structure
• Drafting — first-draft prose generation from outline + data inputs
• Calculator widget logic — math formulas and form validation
• Code infrastructure — site framework, schema markup, database queries

What AI does NOT do unsupervised:

• Set pricing figures (always human-verified against primary sources)
• Decide which contractors or brands to recommend (we follow methodology, not AI opinion)
• Publish without human review (every article goes through tone + accuracy checks before going live)

We disclose AI involvement directly because (a) it's increasingly common in the publishing industry but rarely disclosed, and (b) readers deserve to know how content they're reading was produced. Google's policy specifically permits AI-assisted content when it's genuinely useful and accurate — we believe transparency is the right standard regardless of policy.

5. Conflict of interest policy

CostPatch earns revenue from two sources:

1. Lead-generation affiliate commissions — when readers request quotes through embedded forms (Networx, Bark, Angi, etc.), we earn $5–$50 per qualified lead routed to a contractor in those networks.
2. Display advertising — Google AdSense initially, Mediavine after we cross 50,000 monthly sessions.

Our editorial direction prioritizes accurate cost ranges even when they reduce lead-gen revenue. Specifically:

• We publish ranges that often steer readers AWAY from full quotes (e.g. "you don't need foundation underpinning; a $500 crack repair suffices")
• We disclose contractor tactics that drive up quotes (sales pressure, financing-driven pricing, scope inflation)
• We name brands and price tiers transparently — including when sales-organization installers charge 30–50% above local independents

If you suspect editorial bias on a specific article, please email [email protected] with the specific section and we'll review.

6. Corrections policy

When we get something wrong, we fix it visibly:

• Factual errors (wrong price, wrong contractor tactic claim, wrong building code reference) — corrected within 7 days of confirmation, with a "Correction" note added to the article footer for 90 days.
• Material-cost changes in our data — refreshed quarterly without a "correction" note since these are expected updates, not errors.
• Reader-submitted ground-truth divergence — if 3+ readers report quotes materially outside our state range for the same service, we trigger a contractor survey and update if confirmed.

All correction submissions can be sent to [email protected].

7. What we don't publish

• Specific contractor recommendations — we route to lead-gen networks, not individual contractors, to avoid pay-to-play bias
• Sponsored content — we do not accept payment for editorial placement. Lead-gen referral commissions are post-conversion, not pre-publication.
• YMYL professional advice — we don't tell you whether your specific foundation needs piers, only what the typical price range is when piers are spec'd. Always get a structural engineer's report before signing.

8. Author transparency

Every article on CostPatch is attributed to "CostPatch Research" — our editorial team, not an individual byline, because the workflow combines AI research, human verification, and group review. Our team backgrounds are on the team page.