seo-local Checked

1. **Validate target & preflight.** See skills/seo-firecrawl/references/preflight.md for the canonical 3-stage preflight (credit balance, Firecrawl availability, Google APIs). Skill-specific notes:

• Normalise the target (strip protocol from domain; confirm homepage is fetchable). Confirm at least one local keyword was provided — if none, infer from <title> + <h1> of the homepage; if still ambiguous, ask the user before continuing.
• Estimated SE Ranking cost for this skill: ~15–25 credits (1 audit re-check + 3–5 SERP queries + 1 domain overview).
• Firecrawl: optional with WebFetch fallback, ~6–9 Firecrawl credits if available (hard cap 12). When available, steps 4 (GBP-on-page audit), 5 (NAP extraction), and 7 (review scraping) run on the homepage + 5 sample pages + provided review URLs. Without Firecrawl those steps degrade to WebFetch-only — schema/JSON-LD detection and tel: / address element extraction become best-effort prose inspection. Pass --no-firecrawl to force WebFetch-only.
• Google APIs: tier 1 (GSC) unlocks step 8b (GSC local query performance) after the local-pack rank step; tier 2 (GA4) additionally unlocks step 8c (GA4 organic-by-landing-page enrichment). See skills/seo-google/references/cross-skill-integration.md for the full enrichment contract.

2. **Business-type detection**

• Read homepage + /contact + footer prose (WebFetch markdown is enough for this).
• Classify as one of:
• **Brick-and-Mortar** — visible street address, "Visit us at", embedded Maps iframe.
• **Service Area Business (SAB)** — no street address, "serving {region}", "we come to you", areaServed in schema without address.streetAddress.
• **Hybrid** — both signals present (e.g. showroom + service area).
• This determines which checks apply downstream. SAB skips embedded-map and physical-address consistency. Record in LOCAL-SEO-REPORT.md "Snapshot".

3. **Industry-vertical detection**

• From URL patterns (/menu, /practice-areas, /listings, /inventory), <title>, page prose, infer one of: Restaurant / Healthcare / Legal / Home Services / Real Estate / Automotive / Generic.
• This routes citation-source recommendations and schema-subtype recommendations later — load references/local-citation-sources.md for the vertical's Tier-1 directories.

4. **GBP signals on the page** mcp__firecrawl-mcp__firecrawl_scrape (with formats: ["rawHtml"])

• Scrape homepage + /contact (or whichever page has the most local intent).
• From rawHtml extract:
• Embedded Google Maps iframe (<iframe src="https://www.google.com/maps/embed?...">) — record place ID if present.
• Reviews widget / GBP rich snippet markup.
• aggregateRating JSON-LD block (presence is the strongest signal that the site wants stars in SERPs).
• Business hours visibility on page (open-at-search-time correlates with rank — Whitespark's #5 factor).
• Click-to-call: count of <a href="tel:..."> elements.
• GBP profile link: any <a href> to https://g.page/... or https://maps.app.goo.gl/... or https://www.google.com/maps/place/....
• **If Firecrawl unavailable:** WebFetch markdown can detect a tel: link in some renderings but loses iframes and JSON-LD. Mark Maps embed / aggregateRating / GBP profile-link detection as (skipped — Firecrawl required).

5. **NAP consistency** mcp__firecrawl-mcp__firecrawl_scrape on homepage + 5 sample pages

• Sample pages: homepage, /contact, /about, plus 2 service or location pages (pick from sitemap or top traffic pages).
• For each, extract:
• **Visible NAP from rendered prose.** Address pattern (street + city + region + postal), phone (tel: href + display format), business name (logo alt, footer, schema name).
• **NAP from JSON-LD.** Parse every <script type="application/ld+json"> block. Pull name, address.streetAddress, address.addressLocality, address.addressRegion, address.postalCode, telephone.
• **Compare across the 6 page samples + schema.** Any divergence (different phone format on the contact page vs homepage; "Suite 200" missing from one footer; schema phone in international format while page shows local format) → record in nap-inconsistencies.csv.
• **Brick-and-mortar only:** if a Maps iframe is present, attempt to read the embedded address from the iframe URL (the place ID and address are URL-encoded). Compare to page/schema NAP. SAB skips this.
• If nap-inconsistencies.csv is empty after the scan, write nap-inconsistencies.csv as a one-line file with header only and note "NAP consistent across {n} pages and schema" in LOCAL-SEO-REPORT.md.

6. **Local-pack rank tracking** DATA_getSerpResults with country/region filters

• For each user-provided local keyword (or the 1–3 inferred from homepage):
• Call DATA_getSerpResults with the user's country and the most specific region/city the API supports (use DATA_getSerpLocations first to confirm a valid location code if the user supplied a city).
• Capture: top 10 organic, local-pack presence (yes/no), the 3 businesses in the local pack if shown (name, rating, review count), AIO presence.
• Cross-check: is the target domain in the top 10 organic? Is the target business name in the local pack?
• **Save the parsed result per keyword to local-keywords.csv** (columns: keyword,country,location,local_pack_present,target_in_pack,target_pack_position,target_organic_position,top_pack_competitor_1,top_pack_competitor_2,top_pack_competitor_3).
• Note the local-pack-ads caveat: the SE Ranking SERP returns the AI/ads-modified pack as Google serves it. If the local pack shows ads, record that — local-pack ad density jumped from 1% to 22% of mobile US local searches in 2025–2026 per Sterling Sky.

7. **Reviews scraping** mcp__firecrawl-mcp__firecrawl_scrape on user-provided review URLs

• **Inputs (user-provided, optional).** GBP listing URL (https://www.google.com/maps/place/...), Yelp business URL, Trustpilot business URL, BBB profile URL.
• **For each provided URL:** scrape with formats: ["rawHtml"]. From the parsed DOM, extract: total review count, average rating, date of most recent review (review velocity proxy), count of owner responses on the most recent 10 reviews.
• **Aggregate signals:**
• Velocity: ≥1 new review in last 18 days = healthy (Sterling Sky 18-day rule). >21 days since last = "review cliff" risk.
• Volume: <10 Google reviews flags below the magic threshold.
• Star rating: 4.5+ matches c