(Insight) GEO · 11 min · 2026-10-26
AI Visibility Tools: What They Measure
AI visibility tools record whether a brand, product, or URL appears in generative answers, and whether that appearance is a bare mention or a citation with a source. An AI visibility platform keeps a stored prompt set, a refresh cadence, and exportable rows so the next run can be compared with the last. The provider decides what to show, and neither a mention nor a citation is guaranteed.

What AI visibility tools do
An AI visibility tool records what a generative answer did with a brand. A mention is the brand, product, or domain named in the answer. A citation is a source the answer attributes: a named organization tied to a claim, or a URL a reader can open. OpenAI's help page on web search says a ChatGPT answer that searched the web may show citations, that those citations can be incomplete, outdated, or incorrect, and that placement is not guaranteed.
Buyers use that record to see which prompts name the brand, which link a URL, and whether the record moved after a page shipped. Search engines and AI providers decide what to show.
Google's AI answers are one system among others. Google's guide to AI features says the SEO best practices still apply to AI Overviews and AI Mode, that there are no additional requirements to appear, and that a supporting link has to be indexed and eligible to show a snippet. The same guide says meeting the requirements does not mean Google will crawl, index, or serve the page. The optimization guide says these features use Google's core Search ranking and quality systems, and that, from Google Search's perspective, this work is still SEO. How to rank in AI Overviews keeps those limits on one page. A ChatGPT citation is a separate system. So is a link in Perplexity. Perplexity's crawler documentation says PerplexityBot surfaces and links sites in Perplexity search results. Allowing a crawler makes a fetch possible. A citation is a later observation from a stored answer.
An AI visibility platform stores the prompt set, the engine, the date, and the answer so the next run can be compared with the last.
AI brand visibility tools point that record at a brand rather than at a single URL. The export still has to separate three outcomes: the brand was named, a URL on that brand's domain was linked, or a third-party page was the source the answer linked. A model can name a brand and link someone else's article. Both facts belong on the row.
Core metrics
These names are category language for any export that stores the answer. A figure is a measurement when two people can score the same stored answer the same way. The labels this site uses for a run, including hard citations, soft citations, competitor citations, and unattributed answers, are on How to measure AI citations.
Mention rate
Mention rate is the share of stored answers, in a defined prompt set, that contain a brand, product, or domain string listed before the run. The denominator is the number of answers stored. A mention does not require a link. Write the strings first: the legal name, the product name, and any misspelling you will count. A comparison page that contains the name is a mention under a "the string appeared" rule. Who was cited as the source is a separate column.
Citation share
Citation share is your linked sources divided by the linked sources in the same prompt set. SEO vs AEO vs GEO describes the related GEO measurement as citation presence, the cited URL, and share of voice on a stored query set. Share of voice sets your citations against the other tracked sources in the panel.
Name the denominator. The share of your own URLs on the homepage is concentration. Citation share uses every linked source in the set. Store the full URL, then roll hosts up. Keep each engine on its own rows: ChatGPT, Perplexity, Google AI Overviews, AI Mode, Gemini, and Copilot do not share one citation object. The export is one row per prompt, per engine, and per date.
Sentiment
Sentiment is a label for the wording around a mention. A usable rubric says what counts as favorable, neutral, or unfavorable, and it quotes the sentence. Keep that label in a different column from the citation. A figure with no sentence and no rubric is an opinion. Misattribution belongs on the same row: the answer names you for a claim you do not make, or names someone else for a claim that originated with you.
Prompt coverage
Prompt coverage is which questions were asked, and in which engines. A long list can still miss the category question, the comparison, and the branded question. Three wordings of one intent show whether a citation survives paraphrase. Label intent before the run. If Gemini or Copilot is absent from the export, that run did not measure them. Ask for the named engines and whether the answer was search-backed.
Source domains
Source domains are the hosts an answer linked, taken from the full URL. Split your domain, another brand's domain, and other hosts such as publishers and forums. A publisher citation is a citation of that publisher, even when the brand is the subject of the sentence. Keep the path, so the row shows the page that was linked. AI brand visibility tools that stop at the domain hide that page.
| Record | Denominator | What the row includes |
|---|---|---|
| Mention rate | Stored answers in the prompt set | The brand strings listed before the run |
| Citation share | Linked sources in that same set | Your URLs and the other linked URLs |
| Sentiment | A written rubric on the stored wording | The sentence that was labeled |
| Prompt coverage | Intents and engines listed in advance | The prompt text and the engine |
| Source domains | Hosts and URLs the answer linked | The full URL, then the host |
An evaluation checklist
Five questions apply to any AI visibility platform. The artifact is a stored prompt, a named engine, a date, and a row you can open.
Prompt sets
The set is written down before the run and labeled by intent, including paraphrases. Each scored prompt starts a fresh thread, because a follow-up inherits the earlier turn. The stored field is the prompt text.
Engines covered
Coverage is a list. ChatGPT, Perplexity, Google AI Overviews, AI Mode, Gemini, and Copilot are separate products. Google's Generative AI performance report reports impressions from AI Overviews and AI Mode. The help page says an impression is a link shown in a generative AI feature, and that the report can be missing when those impressions are too few. Google's AI features guide says these appearances are also counted in the Performance report under the Web search type. That report is the Google record. A ChatGPT observation stays on a ChatGPT row.
Refresh cadence
A cadence repeats the same prompts and stores the date, the interval, and the time zone. A comparison is two stored runs of the same prompt, with the interface name on the row when the product shows it.
Exportable data
Keep the prompt, the engine, the date, the mention, the citation, the linked URLs, and the other brands named. A colleague should be able to score one answer from the stored text.
Change tracking
Change tracking ties a before and an after to a URL that shipped. Store the baseline, write the verification window before deploy, and report improved, no change, or regressed. SEO change tracking does that for Search Console and crawl evidence on one isolated domain. A prompt panel follows the same order: store the run, ship the page, run the same prompts again.
Categories of AI visibility tools
These are jobs. This page does not rank vendors or list their prices.
Prompt and citation panels. They run a fixed prompt set and store mentions, citations, and linked URLs. Judge them with the checklist: frozen prompts, named engines, a refresh, an export, and a previous row.
Provider reports. The engine publishes numbers for its own product. For Google, that is the Generative AI performance report, plus appearances counted in the Performance report under the Web search type, as the AI features guide describes. Keep that report in the Google column.
Crawl and eligibility checks. They read robots.txt, noindex, and sometimes a sitemap or an llms.txt file, and they answer whether a crawler may fetch a URL. What is llms.txt? is a proposed Markdown file that maps a site for an agent. The optimization guide says Google Search ignores llms.txt, so the file neither helps nor harms visibility or rankings in Google Search. A crawler check is the eligibility question. Citation share is the stored-answer question.
Change-tracking control planes. They bind evidence to one domain, name the allowed change, require a live URL before the task closes, and compare a declared window with a baseline. They leave the choice of what to show with the answer engine.
A crawler report answers a fetch. A provider report answers that provider's impressions. A prompt panel answers the stored questions. A control plane answers whether the shipped page is the page the work order named.
How to measure AI visibility yourself
A second person should be able to follow the protocol.
- List the brand strings, the product names, and the canonical hosts you will count. List the other brands in the denominator the same way.
- Write a short prompt set, label each prompt's intent, and include paraphrases. Include one branded prompt and one unbranded comparison prompt.
- Open a fresh thread for each prompt. Record the engine, the date, the full answer, and every visible URL.
- Mark a mention when a listed string appears. Mark a citation when the answer attributes a claim to you or links a URL you listed. Keep the marks in separate columns.
- Run the same prompts again. A citation that appears once needs a second stored run before you treat it as stable.
- For Google AI Overviews and AI Mode, read the Generative AI performance report and keep that impression count in the Google column.
One screenshot is a single observation. The longer protocol is How to measure AI citations. What does GEO stand for? is the name this site uses for citability. Ordinary search eligibility is the base: What does SEO stand for?. One page as the direct answer to one question is a different unit: What is AEO?.
How iSEOup approaches AI visibility
iSEOup is the SEO, AEO, and GEO control plane for one isolated domain. Generative engine optimization software is that control plane for GEO. It stores AI-visibility prompts. AI visibility is tracked as prompts, mentions, and citations. Launch includes 100 tracked AI prompts per month. Growth includes 2,000 tracked AI prompts per month. Agency includes 10,000 tracked AI prompts per month. Paid market data stays inside the plan allowance. iSEOup does not publish a citation percentage.
The agent researches demand, priorities, content gaps, entities, and stored AI-visibility prompts from Search Console, analytics, crawl, brand, CMS, and deployment sources inside one project. You connect an MCP client you already use: Cursor, Codex, Claude, Grok, ChatGPT, or another MCP client. The credential is scoped to one project and cannot reach another customer's data. iSEOup does not pay for that agent or for model usage. Connecting ChatGPT as a client does not place a page in ChatGPT search.
AEO software is the same control plane for answer selection: a question, a canonical URL, and an extractable answer. GEO content on that plane is entity clarity and quotable facts. The order is eligibility, then selectability, then citability. The reading rooms are GEO insights and AEO insights.
A governed work order names the target URL, allowed actions, acceptance tests, and rollback conditions. Manual mode waits for approval. Supervised mode stops at every risky decision. Autonomous mode repeats only low-risk work inside explicit limits. Indexability, canonical URLs, internal links, titles, structured data, and rendering issues can become work orders when Search Console and crawl evidence support them. Governed content work orders stay within plan limits: 8 a month on Launch, 30 on Growth, and 100 on Agency. Design, DNS, checkout, and credentials stay out of scope unless the order says otherwise. Third-party placements are never purchased automatically.
The task does not close because an agent says it is done. The production receipt stores the live URL and implementation reference. SEO change tracking stores a Search Console and crawl baseline, then compares the declared measurement window with that baseline. The report is improved, no change, or regressed. Search engines and AI providers decide what to show. Rankings and AI citations are never guaranteed.
The free AI crawler checker answers whether ten AI crawlers may read the site. It reads robots.txt, including the groups for OAI-SearchBot, GPTBot, and PerplexityBot, and it checks noindex, llms.txt, and whether a sitemap is reachable. That check is separate from prompts, mentions, and citations.
Agency includes client reporting and branded exports. The stored AI-visibility record described on the product pages remains prompts, mentions, and citations.
Launch is $49 per month for one isolated domain. Growth is $199 per month for one isolated domain. Agency is $699 per month for five isolated domains. There is no free trial. Subscriptions are billed when checkout is completed. The same prices are on pricing.
FAQ
What are AI visibility tools?
AI visibility tools record whether a brand, product, or URL appears in generative answers, and whether that appearance is a mention or a citation. The unit on this site is a stored prompt set and the next run of the same prompts. Search engines and AI providers decide what to show. Placement is not guaranteed.
What is an AI visibility platform?
An AI visibility platform keeps a prompt set, the engines it runs, a refresh cadence, and rows you can export, so a later run can be compared with an earlier one. iSEOup tracks AI visibility as prompts, mentions, and citations inside one isolated project (https://iseoup.com/geo-software). It does not publish a citation percentage.
What do AI brand visibility tools measure?
The category records are mention rate, citation share, sentiment, prompt coverage, and source domains. Mention rate counts stored answers that contain the brand string. Citation share is your linked sources among the linked sources in the same set. Sentiment is a rubric label for the wording around a mention. Prompt coverage is which questions and engines were included. Source domains are the hosts the answer linked. Those are observations from a stored run.
How do you measure AI visibility yourself?
Freeze a prompt set, run each prompt in a fresh thread, and store the date, the engine, the full answer, and the linked URLs. Repeat the same prompts and compare the next run (https://iseoup.com/insights/measure-ai-citations). For Google, the Generative AI performance report in Search Console includes impressions from AI Overviews and AI Mode. An impression is a link shown in a generative AI feature (https://support.google.com/webmasters/answer/16984139). The report can be missing when those impressions are too few.
Does iSEOup guarantee an AI mention or citation?
No. Search engines and AI providers decide what to show. iSEOup records a baseline, stores the prompts and the production receipt, and reports improved, no change, or regressed. Launch is $49 per month for one domain, Growth is $199 per month for one domain, and Agency is $699 per month for five isolated domains.
Is an AI Overview impression the same record as a ChatGPT citation?
No. An AI Overview is a Google Search feature. Google's guide says there are no additional requirements to appear in AI Overviews or AI Mode beyond a page that is indexed and eligible for a snippet, and a site included in Search generative AI features. Meeting them does not mean Google will crawl, index, or serve the page (https://developers.google.com/search/docs/appearance/ai-features). A ChatGPT citation is a separate system, and OpenAI says placement is not guaranteed (https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt).