Where does your company sit?
Type your company name and website once. The self-check uses them on this page and in your re-test link, and nowhere else: nothing is sent to us or stored on our servers. Then answer twelve questions honestly. “Not sure” is a real answer and scores the same as “no”, because an engine cannot use what you cannot confirm.
Your company’s shortlist score
Band
Score out of 100
Area by area
Fix first
Your self-check covers what you can see. What ChatGPT, Perplexity and Google’s AI actually say about your company needs the free AI Visibility Snapshot: a sample of your buyers’ questions on the main engines, human-reviewed, in your inbox within 24 hours.
Get your free AI Visibility Snapshot ↗Keep this result
This link holds your answers, so you can re-test next month and see what changed. Nothing is stored on our side.
Save my result and get the committee PDF ↓ Use the short form at the end of this page; your re-test link is added to the message for you.
Looking for a raw count of how often AI tools mention your brand? Use the free checkers from Ahrefs and Semrush; both work without an account, as of 8 November 2026. They count mentions. This self-check measures something different: whether your company has what an engine needs to put you on a shortlist, and whether you could prove the result to your board.
How the score works.
Each of the six areas carries a weight, and each area has two questions of equal value. The first answer to every question earns full marks, the second earns half, and “no” or “not sure” earn nothing. The total is out of 100. Areas that hold the most weight are the ones the published evidence ties most closely to being named: what your own site states, and the proof an engine can quote.
| Area | Weight | What it asks |
|---|---|---|
| 1. Facts on your own site | 20 | Can an engine read who you are, what you do, for whom and where? |
| 2. Profiles the engines read | 15 | Are your review and directory profiles complete and consistent? |
| 3. Proof engines can quote | 20 | Are your results specific, dated and on readable pages? |
| 4. Pages the rewritten searches retrieve | 15 | Are you on the comparison and list pages the engines’ searches return? |
| 5. Access for AI crawlers | 15 | Can the engines’ crawlers fetch and read your pages? |
| 6. Measurement to pipeline | 15 | Can you see and report what AI answers send you? |
The five bands
| Score | Band | What it usually means |
|---|---|---|
| 0 to 24 | Hard for AI to find | Basic facts or access are missing. Start there before anything else. |
| 25 to 44 | Visible in parts | Some pieces exist, but an engine would struggle to describe you with confidence. |
| 45 to 64 | Partly shortlist-ready | The foundations are there; proof, profiles or list coverage are holding you back. |
| 65 to 84 | Shortlist-ready | You have removed most of the reasons an engine could not name you. |
| 85 to 100 | Strong shortlist candidate | The remaining work is testing, earning more independent mentions, and measuring. |
A high score does not mean an engine will name you. Nobody outside the engines controls that, and anyone who promises it is guessing. It means you have removed the reasons it could not. When two areas score the same, the “fix first” list puts the earlier link in the chain first: access, then facts, then proof, profiles, retrieved pages and measurement.
Facts on your own site.
When an engine searches for suppliers, it reads pages and extracts which companies they describe. Your own site is the one source you fully control, so it should say plainly what you do, which buyers you serve, in which countries, and since when. A home page that opens with a slogan, and an about page that tells a founding story without the facts, leave the engine to guess or to borrow from older copies elsewhere.
Google’s documentation for company structured data recommends placing company details “on your home page, or a single page that describes” the company, and lists properties such as founding date, address, number of employees and links to your other profiles (accessed 7 November 2026). Google also says no special markup is needed to appear in its AI features, so treat structured data as a tidy summary of what the page already says, never a replacement for it.
What good looks like
One sentence on the home page that a stranger could repeat: what you do, for whom, where. One facts page with founding year, location, team size, services, sectors and the profiles you own. The same facts, in the same words, everywhere. The play Do we need a company facts page for AI engines to read? walks through it step by step.
Profiles the engines read.
A shortlist question needs pages that compare firms, and review sites and directories do exactly that. In a G2 survey of more than 1,000 B2B software buyers published on 6 October 2025, half said they now start their research in an AI chatbot rather than Google, and AI chat ranked first among the sources buyers use to build a shortlist, with review websites second. In Profound’s study of AI citations (August 2024 to June 2025, updated August 2025), G2 was among the five domains ChatGPT cited most, and LinkedIn among the five that Google’s AI Overviews and Perplexity cited most.
A profile that exists but is thin, or carries reviews from three years ago, tells an engine little. A profile whose founding year or headline result differs from your site gives it two answers to choose between.
What good looks like
Complete profiles on the two or three directories that actually rank for your category searches, with genuine client reviews from the last 12 months, and facts that match your own site line by line. Never buy or script reviews: review sites remove them, and in August 2024 the US Federal Trade Commission announced a final rule banning fake reviews and testimonials. The plays Which review directories do AI engines read for our category? and Do our facts match across every profile the engines read? cover the method.
Proof engines can quote.
An engine asked “who is good at this?” looks for evidence it can repeat. A specific, dated result with a named source (“organic sign-ups rose from X to Y between these dates, measured in this tool”) gives it something firm. “We deliver outstanding results” gives it nothing. Google’s own guidance for its AI features, published on 21 May 2025, asks site owners to “focus on making unique, non-commodity content”, and your client results are the least commodity content you have.
Freshness matters too. A Seer Interactive study published on 24 July 2026 found that 75% of pages cited in AI answers had been updated within the previous 12 months. Results locked in a PDF behind a form, or in a sales deck, are invisible to a crawler.
What good looks like
At least one client result on each service page, with the measure, the tool, the dates and the comparison period. Case studies as ordinary web pages, reviewed each year, with the date visible. Client names only where the client has agreed in writing. Read How do we write case studies that AI engines can quote? and How do we make our results quotable, with units and dates?
Pages the rewritten searches retrieve.
Engines do not search for your buyer’s exact sentence. They rewrite it. In a Grow & Convert study of 100 buying-intent prompts (last updated 30 March 2026), 83 triggered a web search, each generating two to four shorter “fan-out” searches. Google describes the same technique for AI Overviews and AI Mode as “issuing multiple related searches across subtopics and data sources”. You are competing for the pages those shorter searches return.
For shortlist questions, those pages are mostly lists. Ahrefs found that “best X” list articles were 43.8% of the page types ChatGPT cited across 750 recommendation prompts (4 December 2025). AirOps reported on 12 March 2026 that pages whose titles closely matched the query were cited at 20.1%, against 9.3% for pages that did not. Both studies come from companies that sell search tools, and both describe what was observed, not how the engines are built.
What good looks like
A written list of the shorter searches behind your buyers’ questions, and a monthly look at what ranks for them. Honest comparison and “best for” pages on your own site, titled as the question a buyer asks. A place on the independent lists that rank, earned through complete profiles, real reviews and editors who have a reason to include you. Read Do we have a page that answers the “best X for Y” searches AI runs for our category? and How do we get into “best agencies” roundup lists honestly?
Access for AI crawlers.
This is the cheapest area to fix and the most expensive to miss. OpenAI documents separate crawlers: OAI-SearchBot “is used to surface websites in search results in ChatGPT’s search features”, and “sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers” (accessed 7 November 2026). Perplexity documents PerplexityBot as the crawler that surfaces and links websites in its search results. Google asks that crawling be allowed “in robots.txt, and by any CDN or hosting infrastructure”.
The trap is usually not robots.txt but a firewall or bot-protection rule that blocks unfamiliar crawlers by default, or pages whose main text only appears after scripts run or a cookie banner is dismissed.
What good looks like
Open yourdomain.com/robots.txt and search for OAI-SearchBot and PerplexityBot. Ask whoever runs your CDN or firewall whether those crawlers are allowed. View the source of a key service page and check the main text is there. Then read Can AI crawlers actually reach our website? and Can AI engines read our content without running JavaScript?
Measurement to pipeline.
The other five areas decide whether you can be named. This one decides whether anyone inside your company will keep paying for the work. GA4’s default channel group has no AI channel; Google’s own help page shows how to build a custom channel called “AI assistants” with a pattern that matches the assistants’ referral sources, and notes that custom channel groups apply to past data too (accessed 7 November 2026).
Visits are only the start. Forrester’s State of Business Buying 2026 (21 January 2026) describes buying groups with 13 people inside the buyer and 9 outside, and says 94% of buyers use AI during the purchase. A committee that size needs more than a traffic chart. It needs to see enquiries and deals that came from AI answers, recorded in the CRM, and reported apart from website behaviour and from visibility itself.
What good looks like
An “AI assistants” channel in GA4. A source field in the CRM, filled from the form and from a “how did you hear about us?” question. A monthly one-page report that keeps visibility, website behaviour and pipeline in separate rows. Start with How do we track AI referrals in GA4? and How do we connect AI-referred leads to closed revenue?, then our guide to reporting AI visibility to your board.
What a self-check cannot see.
This scorecard measures what you can check from the inside. It cannot tell you what the engines actually say, and one test of your own would not settle it either. Research by SparkToro with Gumshoe.ai, reported by Search Engine Journal on 30 January 2026, ran 2,961 prompts through ChatGPT, Claude and Google’s AI and found less than a one-in-100 chance of getting the same list of brands twice. The researchers concluded that how often a brand appears across many runs is more consistent than its position in any one answer.
There is also a gap between what is read and what is shown. OpenAI’s documentation for its web search tool notes that “the number of sources is often greater than the number of citations”: a page can shape an answer without appearing as a link under it. Measuring that properly takes a fixed set of buyer questions, several runs per engine, every company named recorded, and a person reading the answers.
Questions buyers ask.
Is this an AI visibility checker?
No. Free checkers from Ahrefs and Semrush count how often AI tools mention a brand. This self-check scores whether your company has what an engine needs to put you on a shortlist, and whether you can measure the result. Use both: a mention count tells you where you are, this tells you what to fix.
Does a high score mean ChatGPT will recommend us?
No. Nobody outside OpenAI, Google or Perplexity controls what their engines say, and answers change from run to run. A high score means you have removed the common reasons an engine could not name you. Treat any promise of placement in AI answers with caution.
What happens to the answers I enter?
They stay in your browser and in the re-test link you can copy. Nothing is sent to our servers. If analytics is enabled, we record only that the self-check started and which of the five score bands it ended in, never your company name, website or answers.
How often should we re-test?
Monthly, or after any change you make from the fix-first list. Open your saved link, retake the twelve questions, and the result shows how your score moved since last time.
Why does measurement count if it does not change what AI says?
Because AI visibility work that cannot show its value rarely survives the next budget round. Measurement does not get you named; it decides whether you can keep investing in the five areas that do.
What does the free AI Visibility Snapshot add?
What the engines actually say. We run a sample of your buyers’ questions across ChatGPT, Perplexity and Google’s AI and show who gets named, which sources are cited, and whether your company appears. A person reviews every Snapshot, it is free, and it arrives within 24 hours with no call required.
Sources: Google Search Central, company structured data and AI features documentation (accessed 7 Nov 2026); Google Search Central blog, 21 May 2025; G2 buyer survey, 6 Oct 2025; Profound citation study, updated Aug 2025; FTC final rule on fake reviews, Aug 2024; Seer Interactive, 24 Jul 2026; Grow & Convert, updated 30 Mar 2026; Ahrefs, 4 Dec 2025; AirOps, 12 Mar 2026; OpenAI crawler and web search documentation (accessed 7 Nov 2026); Perplexity crawler documentation (accessed 7 Nov 2026); Google Analytics help, custom channel groups (accessed 7 Nov 2026); Forrester, 21 Jan 2026; Search Engine Journal on SparkToro and Gumshoe.ai, 30 Jan 2026; Ahrefs and Semrush free checkers (read 8 Nov 2026). Weights and bands: Prolinkage AI Shortlist Scorecard, Edition 1, 8 Nov 2026.