listing-accuracy-audit

Listing Accuracy Audit

You are a local-listing auditor. A wrong category, a stale phone number, or a missing address quietly suppresses local-pack rank and sends ready-to-buy customers to a competitor. This skill reads a business's public map/local listing field by field and flags where the details are inconsistent, incomplete, or off from what they should be — read-only.

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

Use UnifAPI for live evidence

You cannot audit a listing from memory — you have to read the live record exactly as a customer sees it, field by field, and confirm it actually surfaces. Use the unifapi skill to connect (OAuth MCP), then call:

• The listing record — maps/search, local/search — read the public listing field by field: name, address, category (primary + secondary), phone, website, hours, rating, review_count, and place_id. Resolve to a single place_id first so the audit targets one canonical record. Loop the query (the business name from a few nearby search points) to catch duplicate pins — more than one distinct place_id for the same real business is itself a Critical finding.
• Discoverability check — seo/serp — run the business's own name + city query and confirm the listing/site actually surfaces. A wrong, suppressed, or missing listing won't appear for its own name — and that outranks every field-level finding, because no field is worth fixing on a record customers can't find. seo/serp also exposes competing pins occupying the brand query.

UnifAPI reads public data only — it never edits or claims the listing. Keep any billing metadata so the output can state record cost.

Workflow

1. Establish the source of truth — required. Get the business's correct name, address, primary + secondary category, phone, website, and hours from the operator (or its website). Without a source of truth there is nothing to audit against; ask before pulling data. (Read .agents/product-marketing.md / .claude/product-marketing.md first if it exists.)
2. Pull the public listing via maps/search / local/search as it appears on the map/local result. Capture place_id; loop the business name across a couple of nearby search points — if more than one place_id resolves to the business, that duplication is a high-impact finding.
3. Compare field by field against the checklist below, assigning each field a status (pass / fix / gap) and an impact rating (critical / high / medium / low) using the rubric.
4. Confirm discoverability with seo/serp: does the listing surface for the business's own name + city? If it doesn't, that finding leads the report — it outranks every field-level issue.
5. Order by impact. Sort findings by impact rating so the operator fixes the rank-and-trust killers first, not the cosmetics.