AI Visibility Tester

Measure how AI answer engines recommend your business — and get a prioritized, evidence-backed list of what's holding your visibility back.

Customers are increasingly asking AI assistants questions like:

• "What are the best dental clinics in Lagos?"
• "Where can I get affordable teeth whitening?"
• "Which accounting firm should I hire?"
• "Best restaurants near me?"

AI Visibility Tester shows whether your business appears in those answers, how AI positions you against competitors — and, critically, why it doesn't, with a fix list traceable back to real evidence.

Instead of measuring traditional search rankings, this Actor measures:

When customers ask AI who they should choose, does your business appear — and what's stopping it from appearing more?

What this Actor does

Input is deliberately minimal — everything else is discovered automatically:

• You provide: business name, website, and location.
• The Actor discovers: niche, services, target customers, and unique value propositions (by crawling and extracting from your actual site — never guessed or inferred from "typical" businesses in your category), then generates realistic customer questions across three intents (transactional, commercial investigation, informational), tests them against AI engines, and identifies competitors purely from who else AI mentions in the same responses.

There is no manual question list or competitor list to fill in. This is intentional: manually-declared competitor lists get stale, and hand-written questions rarely match how real customers actually phrase things.

The Actor then measures:

• Business mentions, recommendations, and (where explicitly given) ranking position
• Website citations, verified against your actual domain, not self-reported
• Competitors AI mentions in the same responses
• Technical AI-crawlability issues (robots.txt blocking, missing structured data, missing meta descriptions)
• Service page coverage — which of your services have no dedicated page for AI to find
• Citation source mix — how much of your AI visibility comes from your own site vs. directories, reviews, or other third parties
• A prioritized, evidence-linked list of opportunities to fix

Example

Question: "What is the best dental clinic in Victoria Island for braces and dental implants?"

Business: Exclusive Smile Dental Clinic

AI response:

For braces and dental implants in Victoria Island, I'd start with Exclusive Smile Dental Clinic. Other options include Richmond Dental Clinic and The Ridge Dental Practice.

Output:

{
  "mentioned": true,
  "recommended": true,
  "position": 1,
  "cited": false,
  "businesses_mentioned": ["Richmond Dental Clinic", "The Ridge Dental Practice"]
}

Interpretation: AI recognizes and recommends the business, places it first when ranking explicitly, but did not directly cite the business website.

Supported AI engines

ChatGPT

Uses OpenAI's web-enabled models to evaluate how ChatGPT responds to customer-style questions. Measures mentions, recommendations, rankings, competitor visibility, and source usage.

Perplexity

Uses Perplexity's search-based AI responses. Measures mentions, recommendations, competitor comparisons, and citations.

Not currently supported: Google AI Overviews and Microsoft Copilot — neither exposes a public API that reliably reproduces its consumer AI search experience.

Fallback mode: if live API access to an engine isn't available, the Actor can fall back to a training-knowledge-only mode for testing purposes. This mode has no live web search and does not reflect real-time AI search behavior — treat any run relying on it as a functional test, not a client-facing visibility result.

Understanding AI visibility

AI visibility is not the same as SEO ranking.

Traditional SEO asks: "Where does my website rank?" AI visibility asks: "When AI recommends businesses, does my business appear?"

A business can be recommended without being cited:

"Exclusive Smile Dental Clinic is a good option for cosmetic dentistry."

{ "mentioned": true, "recommended": true, "cited": false }

The AI knows and recommends the business but doesn't link directly to the website.

Or a business can be cited:

"According to Exclusive Smile Dental Clinic's website..."

{ "cited": true, "citation_url": "https://example.com" }

The business has earned a direct AI source reference.

Output

Each query × AI engine combination produces one visibility result:

{
  "query": "best dental clinics in Lagos",
  "engine": "chatgpt",
  "status": "success",
  "mentioned": true,
  "recommended": true,
  "position": null,
  "cited": false,
  "citation_url": null,
  "businesses_mentioned": ["Richmond Dental Clinic", "Schubbs Dental Clinic"],
  "raw_response_excerpt": "..."
}

Output fields

Field	Description
mentioned	Whether the business appears in the AI response at all.
recommended	Whether the AI presents the business as a suitable choice (e.g. "worth checking," "top choice," "highly regarded").
position	The business's position when the AI gives an explicit ordered ranking (e.g. a numbered list). null if the AI recommends businesses without ordering them.
cited	Whether the AI response cites the business's own website as a source. Verified against the business's actual domain — a directory or aggregator site mentioning the business doesn't count as a citation.
businesses_mentioned	Other businesses AI mentioned in the same response — discovered automatically, not from a declared list. The business's own name/aliases are filtered out and near-duplicate mentions are merged.

Technical & content findings

Beyond testing AI responses, the Actor runs deterministic checks against the crawled site itself — no LLM guessing, every finding traceable to a specific page or check:

{
  "check_id": "robots_ai_crawlers",
  "status": "fail",
  "severity": "high",
  "detail": "robots.txt disallows the following AI crawler(s) from the entire site: GPTBot.",
  "evidence": { "blocked_agents": ["GPTBot"] }
}

Checks include: AI crawler access in robots.txt, structured data (schema.org) presence, missing meta descriptions on key pages, and whether each of your extracted services has a dedicated page.

Citation source analysis

Every citation an AI engine uses is classified by source type — first-party (your own site), directory, review site, social, community, news, or unknown — so you can see, for example, that AI visibility is being carried entirely by directory listings rather than your own content.

Opportunities

The Actor's most client-facing output: a prioritized, evidence-linked list of what to fix, generated from fixed rules against the technical findings, service coverage, citation mix, and per-intent visibility data above — not an LLM guessing at what might be wrong.

{
  "issue": "No dedicated page found for 'Invisalign'.",
  "severity": "high",
  "category": "content",
  "recommendation": "Publish a dedicated page for 'Invisalign' covering cost, process, and eligibility.",
  "estimated_impact": "high"
}

Summary metrics

Metric	Meaning	Example
Mention rate	How often the business appears across all tested queries.	1.0 = mentioned in 100% of tests
Recommendation rate	How often AI recommends the business (not just mentions it).	0.83 = recommended in 83% of tests
Citation rate	How often AI cites the official website as a source.	0.25 = cited in 25% of tests
Average position	Average explicit ranking position, across only the queries where AI gave an ordered list.	1.5

Visibility is also broken down per intent (transactional / commercial investigation / informational) — a business can be strong on informational questions and nearly invisible on transactional ones, and blending those into a single number would hide that.

How it works

Business Name + Website + Location
            |
            v
   Crawl & Extract Business Profile
            |
            v
   Generate Customer Questions (3 intents)
            |
            v
   AI Search Engines (ChatGPT / Perplexity)
            |
            v
   Validation Layer (logic checks, citation
   verification, self-mention filtering)
            |
            v
   Technical Findings + Citation Classification
            |
            v
   Opportunity Detection (rule-based)
            |
            v
   Visibility Report + Prioritized Fix List

Validation & accuracy

AI responses can contain inconsistencies, so every result passes through a validation layer before being returned.

Citation validation — a citation only counts when the URL's domain matches the business's actual website.

• Counts: business website is https://example.com, citation is https://example.com/services
• Doesn't count: citation is https://directory.com/example-business

Logical validation — impossible combinations are corrected automatically. For example, a response claiming:

{ "mentioned": false, "recommended": true }

is invalid — a business can't be recommended without being mentioned — and is corrected before being returned.

Self-mention filtering — the business's own name (and declared aliases) never appear in its own businesses_mentioned list, even when AI response phrasing varies (e.g. "Example Dental" vs. "Example Dental: Dr. Smith").

Use cases

Local businesses — dental clinics, restaurants, lawyers, hotels, real estate companies, healthcare providers — understand whether AI recommends you at all, and exactly what to fix if it doesn't.

AEO agencies — build AI visibility audits as a repeatable, evidence-backed workflow:

Run audit -> Review prioritized opportunities -> Pitch the fix -> Re-run monthly to show progress

Competitor research — discover which businesses AI recommends most, who shows up for your important queries, and where your business is missing.

Ongoing monitoring — every run is archived, so re-running over time shows measurable visibility and citation-mix improvement, not just a one-time snapshot.

Best practices

Let the Actor generate questions — don't test brand searches yourself.

	Example
Avoid	"Exclusive Smile Dental Clinic" (only tests brand recall)
What the Actor generates	"best dentist in Lagos for dental implants"

Because questions are generated from your site's actual extracted services and location, results are automatically specific — no manual question-writing needed.

Limitations

Engine	Status
ChatGPT	Supported (requires API access)
Perplexity	Supported (requires API access)
Google AI Overviews	Not supported — no reliable public API
Microsoft Copilot	Not supported — no reliable public API

Other known limitations:

• The archive dataset doesn't deduplicate re-runs — repeated runs accumulate rows rather than replacing prior ones.
• estimated_impact on opportunities and recommendation text are currently rule/template-based, not LLM-phrased — deliberate, for traceability, at the cost of more generic wording.

Roadmap

• LLM-phrased recommendation text (on top of the current rule-based detection)
• Historical visibility dashboards (the archive dataset already accumulates the data)
• Multi-location monitoring
• Google AI Overviews / Copilot support, if a reliable public API becomes available

Deliberately not planned: a single blended "AI Visibility Score." Multiple transparent, evidence-linked metrics are more defensible than one black-box number — see Summary metrics above.

About

Traditional SEO measures where you rank. AI visibility measures whether AI recommends you — and now, exactly what to do about it.

AI Visibility Tester helps businesses and the agencies that serve them understand and improve their presence in the next generation of search.