AI Visibility Checker for Local Businesses

When someone asks ChatGPT, Gemini or Claude "who's a good plumber in Marietta?", does your business come up?

A growing share of customers don't search, they ask. The AI answers with three to five names. This Actor tells a local business (or the agency that serves it) whether it is one of those names, how often, in what position, who gets recommended instead, and which websites the AIs trust when they decide.

Why this one, when there are others

Most AI-visibility tools let you type any prompt you like. That makes month-over-month numbers meaningless, because you never ask the same thing twice. This Actor is built differently:

• Same questions every month. Prompts are generated deterministically from category + location + batteryVersion. Run it in October and again in November and the score measures the same thing. That is what makes it a report, not a screenshot.
• Built for local businesses. Input is a category and a city, not a keyword list. The questions are the ones real customers ask: "affordable but reliable…", "who should I call…", "which ones should I avoid…".
• Tells you why, not just whether. Every answer's cited websites are collected and ranked. If the AIs keep citing Yelp and Expertise.com and you're not on either, that is your to-do list.
• You only pay for answers. Engine errors, rate limits and empty responses are never billed.
• Cheaper at agency volume. Bring your own OpenAI / Gemini / Anthropic keys and pay a fraction of the per-question price.

A real run

Superior Plumbing, a real plumber in Marietta, GA, 5 questions × 3 engines, September 2026:

{
  "rowType": "summary",
  "brand": "Superior Plumbing",
  "category": "plumber",
  "location": "Marietta, GA",
  "visibilityScore": 47,
  "mentionRate": 0.267,
  "avgRankWhenMentioned": 3,
  "shareOfVoice": 0.138,
  "competitorShareOfVoice": {"Plumb Doctor": 0.241, "Dupree Plumbing": 0.241, "Greenlee Plumbing": 0.138},
  "topCitedSources": [{"source": "expertise.com", "count": 18}, {"source": "benjaminfranklinplumbing.com", "count": 13}, {"source": "plumbdoctorga.com", "count": 12}],
  "engineBreakdown": {
    "gemini":    {"prompts": 5, "mentioned": 3, "mentionRate": 0.6, "avgRank": 1},
    "openai":    {"prompts": 5, "mentioned": 0, "mentionRate": 0.0, "avgRank": null},
    "anthropic": {"prompts": 5, "mentioned": 1, "mentionRate": 0.2, "avgRank": 9}
  }
}

The story in one line: Gemini ranks them #1, ChatGPT has never heard of them, and the AIs are reading Expertise.com. That is the conversation an agency wants to have with a client.

What you get

One row per question × engine:

field	meaning
mentioned	did the answer name the business
rank	position among the businesses named (1 = first)
competitorsNamed	every other business the answer recommended
sentiment	how the answer treated the business
citedSources	websites the answer relied on
promptIntent	best, affordable, emergency, compare, near_me… so you can see where it is weak
answerSnippet	the first 400 characters of what the AI actually said
error	why an answer is missing, if it is (never billed)

Plus one summary row with the Visibility Score (0–100), mention rate, average rank, share of voice vs competitors, per-engine and per-intent breakdowns, and the top cited sources. The summary is also stored under the SUMMARY key in the run's key-value store for chaining into report generators.

Engines

• ChatGPT — OpenAI Responses API with web search (real citations)
• Gemini — Google API with Search grounding (real citations)
• Claude — Anthropic API with web search (real citations)
• Perplexity — coming soon
• Google AI Overviews — coming soon, via a licensed SERP data provider; no scraping

Inputs

brand and category are required. Add location for local businesses; leave it empty for national brands. Optional competitors pins the share-of-voice comparison to names you care about; leave it empty and the Actor ranks whoever the AIs actually recommend.

Bring your own keys: pass byoKeys ({"openai": "sk-…", "gemini": "…", "anthropic": "…"}) and you are billed at the lower BYO per-question price.

Pricing

Pay per event, no subscription. Roughly: 10 questions × 3 engines ≈ $1, a 25-question monthly report ≈ $2.35. Errors are free. Exact per-event prices are on the pricing tab.

Use cases

• Agencies: run it for every local client monthly; the summary row drops straight into a report or a Google Sheet.
• Local businesses: find out which AI assistants ignore you and what sites they cite instead.
• AI agents: enabled for agentic usage; call it with a brand, a category and a city and read SUMMARY.