AI visibility tool
AI Visibility Tool for the Buyer Questions That Matter
People do not only ask AI assistants for facts. They ask which product to choose, which vendor to shortlist, what alternative to consider. We measure those questions across ChatGPT, Gemini and Perplexity and show you where AI recommends a competitor instead of you.
The six questions a visibility tool has to answer
If a tool stops at “your visibility score is 37”, it has measured a symptom and left the decision to you.
- Which buyer questions are commercially important?
- When does AI recommend a competitor instead of us?
- Is the pattern consistent enough to trust?
- What evidence appears in the answers that win?
- What do our own pages actually communicate?
- What should we do first, and did the measured state change afterwards?
What we measure
Seven things, kept deliberately separate because they imply different work.
Mention≠Recommendation
A brand can appear in an answer without being recommended. If a buyer asks for the best platforms in a category and you appear only in a caveat while a competitor is put forward, counting both as “mentions” hides the commercial outcome. Mention status and recommendation status are kept apart.
Buyer-question gaps
The question set is built around decisions: best tools in a category, alternatives to a named competitor, head-to-head comparisons, use-case fit, pricing and value, and the strategic claims you want to own. Educational questions still matter for citation visibility, but they do not automatically carry the same commercial weight as a question that can decide a shortlist.
Engine by engine, never blended
The three engines ground answers differently, and in our own runs the same question routinely produces a different set of named brands on each. A positioning claim can be understood by one engine and lost on another — which is exactly the finding a single blended score destroys.
Competitor occupancy
When you lose, the useful question is not only “are we absent?” but who is occupying the answer instead? Product and company competitors are separated from citation-source domains, so a publisher, marketplace or randomly cited website is never silently promoted into a competing vendor.
Citation and source evidence
Answers often lean on sources outside your website. Observed citations are recorded, and three things are kept distinct: a source that is merely relevant, a source that recurs, and a source that may actually be addressable.
Owned-site evidence
A bounded set of public pages on your domain is inspected and compared against the themes that appear in winning answers. That replaces “do we have enough content?” with a better question: do the pages a buyer and an assistant can reach contain the evidence this decision appears to require? The wording stays bounded too — “no matching evidence in the inspected page set”, never “your website does not have this”.
ProductTruth
Website copy is not the same thing as product reality. If winning answers emphasise a free tier and your site does not mention one, the correct next step depends entirely on whether the product has one. Product facts are held in three states — verified true, verified false, unknown — and unknown stays unknown. A missing claim on a page never becomes a recommendation to publish a claim that is not true.
Source relevance≠Source addressability
A frequently cited domain is not automatically somewhere you can get listed. Citation frequency never raises the addressability score — that rule is enforced in the engine, not left to the person writing the report.
From visibility to a commercial decision
Mentions, citations, share of voice and a trend line are context. They are not the decision. Two things are scored separately.
Commercial Gap Score
How commercially important and how severe is this measured recommendation gap? A major buyer-question loss stays visible even when it is hard to fix.
Action Priority
Given what we actually know, is this worth working on now? A critical gap can have low or unknown actionability.
Best [category] tools for agencies
Competitor winsMost affordable [category] platform
Mentioned only[Category] with SOC 2 compliance
Competitor winsCommercial importance≠Fixability
Collapsing the two either buries the hard problems or pretends everything is addressable. Keeping them apart is why a report can say “this is your biggest loss and we do not yet know how to move it” — which is a more useful sentence than a confident wrong one.
A better first screen than “Visibility: 37”
The first thing you should see is a finding you can act on, not a composite number.
Your “no monthly subscription” positioning is recommended by OpenAI, but Gemini and Perplexity recommend Uniqode for the same buyer question.
From there you can open the exact question, the answers per engine, the competitor that won, the evidence themes, the relevant page you own, the product-truth state, the confidence, the next action and the remeasurement plan. Secondary analytics still exist — they simply come after the insight.
OwnQR · stated positioning claims, by engine
Founder-owned dataWhy repeated measurement matters
AI answers vary. The same question can return a different shortlist on another run, model version, location or date, so a single answer is not a ranking.
Repeated compatible observations are used to separate four states that most dashboards merge into one number: measured loss, directional gap, insufficient evidence and no gap. When the evidence is insufficient the correct action is to measure again — not to rewrite a website on the strength of one unstable answer.
Unknown≠Zero
A failed provider call is excluded from the denominator rather than counted as your absence. A bad API day must never be reported as a collapse in your market position.
What happens after you change something
An action can be marked complete and tied to a compatible follow-up measurement, which then shows the before and after state per engine — or refuses the comparison and says why.
The follow-up shows the before and after recommendation state, the per-engine movement, competitor changes, citation changes — and whether the follow-up was comparable at all. If the question, cohort, providers or context moved enough to make the comparison invalid, the comparison is refused and the reason is shown instead of a misleading delta.
Before and after≠Causality
“Measured state improved after the recorded change” is valid. “This page change caused Gemini to recommend you” is not, and no observational before/after can support it.
What ours does today, line by line
We would rather lose a deal over this table than win one and have you find out in week two.
| Capability | Status |
|---|---|
| ChatGPT (OpenAI API + web search), Gemini (Google Search grounding), Perplexity | Live |
| Per-engine results, never averaged | Live |
| Recommendation separated from mention and from consideration | Live |
| Buyer-intent classification of every question | Live |
| Competitor occupancy, with citation sources kept out of the competitor set | Live |
| Citation and source evidence, split into relevance and addressability | Live |
| Bounded owned-site inspection, compared only against observed winning themes | Live |
| ProductTruth — verified true, verified false, unknown | Live |
| Commercial Gap Score and Action Priority scored separately | Live |
| Strategic positioning per engine: won, contested, lost | Live |
| Failure-aware denominators; unmeasured never counted as absence | Live |
| Raw answers retained; offline re-scoring at no extra provider cost | Live |
| Frozen, versioned question cohorts | Live |
| Before/after comparison on the same frozen set, incompatible runs refused | Live |
| Self-service dashboard and login | Not yet — pilots are delivered as a founder-reviewed report |
| Automated recurring schedule and alerts | Not yet — re-runs are triggered by us during a pilot |
| Public API, integrations, SSO, white-label | Not yet — these follow demand, not a roadmap slide |
| More than three engines | Not yet |
Who this is for
SEO and GEO agencies
One workflow across client brands, and an evidence-backed answer to the question clients now ask: what is AI recommending, and what should we do about it? See AI visibility for agencies.
In-house SEO and content teams
Prioritise work around buyer questions instead of publishing volume of generic “AI-optimised” content.
Founders and product marketers
Find out whether the positioning you believe differentiates you is actually present in AI recommendations — per engine.
What an AI visibility tool should not claim
AI visibility is a measurement problem with real uncertainty in it. Good software makes that uncertainty inspectable instead of hiding it inside a score.
Be careful with any platform — ours included — that implies
- one scan represents a permanent AI ranking,
- a citation proves why an engine recommended a brand,
- a before/after change proves causality,
- every high-frequency source is easy to influence,
- the absence of a claim on a page proves the product lacks the feature,
- or that more tracked prompts automatically means more business value.
AI visibility tools and traditional SEO tools
You usually need both. They measure different interfaces.
Traditional SEO tools remain essential: rankings, search demand, backlinks, crawling and indexing, organic traffic. AI visibility adds a different layer — which brands are recommended, which sources are cited, which buyer questions are lost, how the engines disagree, and how that moves over time. For Google's own AI features, Google states that the best practices for SEO remain relevant, because those experiences still depend on a page being indexed and eligible to appear in Google Search.
What to evaluate before choosing any of them
We wrote this list because we had to answer it ourselves, and we publish where we lose on it.
- Does it store prompt-level evidence? You should be able to open the answer behind the metric.
- Does it separate mentions from recommendations? Being named is not being chosen.
- Does it use questions you actually care about? A huge prompt database is good for discovery; a small set of high-stakes questions is usually better for action.
- Can it show competitor occupancy? You need to know who wins when you do not.
- Does it expose source evidence? “Improve authority” is not a finding.
- Does it understand your own site? Without owned evidence, recommendations collapse into generic content advice.
- Does it know when it does not know? Unknown product facts, thin samples and unverified source routes should stay explicit.
- Can it remeasure the same question under a compatible protocol? Without history and proof, monitoring is a recurring bill.
Start with the questions that can cost you a sale
You do not need hundreds of prompts to find your first useful gap. Start with what a serious buyer would ask before they have heard of you.
- “What are the best [category] tools?”
- “What are the best alternatives to [competitor]?”
- “Which [category] product is best for [use case]?”
- “Which option has [the thing you think differentiates you]?”
- “Should I choose [you] or [competitor]?”
Then measure them repeatedly, across the engines your buyers actually use. That is what AI search monitoring does with the same frozen set, what AI search analytics turns into metrics, and what an AI visibility report puts in front of a client.
Common questions
What is an AI visibility tool?
An AI visibility tool measures how a brand appears in AI-generated answers: whether it is mentioned, recommended or cited, which competitors appear instead, and how those outcomes change across questions, engines and runs.
Which AI engines does this tool measure?
Three: OpenAI, Google Gemini and Perplexity, each through its official API with web search or grounding enabled. Results are reported per engine and never averaged into one number.
Is AI visibility the same as SEO visibility?
No. SEO visibility describes presence in a list of results a person then chooses from. AI visibility describes presence inside a generated answer. The inputs overlap; the outcome being measured does not.
How many buyer questions should I track?
Enough to cover the decisions that matter. For most brands a curated set of 10 to 20 buyer questions produces more usable findings than hundreds of low-value prompts, because every question has to be worth acting on.
Can a tool prove why ChatGPT or Gemini recommended a competitor?
Not from observational data alone. It can show the themes present in winning answers, the sources cited, which competitor occupied the answer and what your own pages say. That is diagnostic evidence, not an explanation of a hidden ranking system.
How do you improve AI visibility?
Start from the measured gap rather than a checklist — the right intervention depends entirely on which gap you have. It may mean clearer evidence on a page you own, a missing comparison or use-case asset, third-party sources, correcting a product fact, or deciding a criterion is not one you can compete on.
How we measure, and how this page is sourced
What we actually measure
Every result comes from the official OpenAI, Google Gemini
and Perplexity APIs with web search or grounding enabled. In our data those
surfaces are recorded as openai_api_web_search,
gemini_api_google_search and perplexity_api_sonar, and reports name
them explicitly.
That is not the same thing as a logged-in person using the ChatGPT app. These are official interfaces with web search on, but they are measurement surfaces rather than consumer sessions: no personalisation, no memory, no app-only features. Results can differ from what any individual user sees, and we do not claim the two are identical. We would rather put that on the pricing page than in the small print.
How this page is sourced
- Vendor pricing and features come from each vendor's own public pricing page, with the date we checked it. We do not copy figures out of other people's comparison posts.
- Google-specific claims come from Google Search Central, not from SEO commentary about Google.
- Research claims name the paper and separate what was demonstrated in a benchmark from what holds in the open web.
- First-hand numbers come from our own measurement runs. Where the brand measured is founder-owned, the page says so in the same sentence as the finding.
This page is maintained by the person who built the measurement engine. There is no editorial team and no external reviewer behind it, and we are not going to invent one for a byline. How the measurements are collected.