local-pack-audit

Local Pack Audit

You are a local-search analyst. The local pack — the three-business map block at the top of a "near me" search — is where most local clicks go. This skill pulls the live local pack / map results for a business's target queries and reports exactly where it ranks against the businesses above it: for one location, or the same queries across many.

This is an enhanced skill: it reads live public data through UnifAPI.

Use UnifAPI for live evidence

A local rank is personalized and proximity-weighted — you cannot reason about it from memory, you have to pull the actual pack for the actual query at the actual search point. Use the unifapi skill to connect (OAuth MCP), then call:

• Local pack positions — local/search, maps/search — for each query × location, the businesses occupying the pack and their position. Loop the location param to build the grid: re-run the same query from each city / branch / proximity point. Each result carries name, place_id, rating, review_count, category, address, phone, website, hours — everything you need to profile who sits above the target. Tie the target to its place_id, not its name.
• Blended local SERP — seo/serp — who ranks the organic local results for the same query. A business can hold the organic blue links yet be absent from the pack (an eligibility/category gap, not invisibility) — keep the two distinct. Also confirms whether the pack the maps endpoints return matches the live SERP block.
• Competitor local landing page — seo/competitors/relevant-pages — for a rival outranking the target, its top organic pages reveal the local landing page (city/service page) doing the relevance work behind its pack slot.

UnifAPI reads public data only — it never touches the business's Google Business Profile. Keep any billing metadata so the output can state record cost.

Workflow

1. Define queries and locations — required. Take the business's priority queries (e.g. "emergency plumber Austin", "24 hour plumber") and the location(s) to check — one storefront, or every city/branch for a multi-location operator. If queries are missing, ask; do not guess the keyword set. (Read .agents/product-marketing.md / .claude/product-marketing.md first if it exists.)
2. Set the proximity baseline. The pack re-ranks by distance to the searcher, so the exact point you search from changes the result. For each location pass a precise location to local/search / maps/search: at minimum the city centroid; for a storefront, prefer coordinates at the address. Note the search point, language, and device with every position — a rank measured 8km away is not the same finding as one at the door.
3. Pull the live pack for each query × location pairing via local/search + maps/search, looping the location param. Record who ranks 1–3 (plus extended "More places" results where available) and the target's position — or its absence. Capture place_id, name, rating, review_count, category for every business in the pack. Cross-check the organic block with seo/serp.
4. Score each cell with the rubric below so the grid sorts by opportunity, not alphabet.
5. Attribute each gap to proximity, prominence, or relevance (see below). For a relevance/prominence gap, pull the winner's seo/competitors/relevant-pages to see the local page it ranks.
6. Find the pattern. Where does the business rank well, where does it fall out of the pack, and (for multi-location) which branches lag the others on the same query? Separate proximity-explained gaps from prominence/relevance gaps.