AI visibility, measured
When someone asks ChatGPT what you do, does it name you?
Beacon asks live AI engines the questions your buyers ask and records whether the answer names your domain. Then it reads your page the way an AI crawler does: no JavaScript, no login.
Free. Three scans a day without an account. A result takes about twenty seconds.
What the engine actually said
Beacon writes eight buyer questions from the page's own description, asks each engine every one, and records two things per answer: whether the engine searched the web, and whether the answer names the domain. Being retrievable and being named are different things; an engine can search, find you, and still not say your name.
Engines that answered in this scan: GLM (Zhipu), ChatGPT. Not connected when it ran: Perplexity, Claude; each switches on when a key is configured. Every question carries the suffix “Answer in under 120 words, naming specific products.”
“Where can designers and developers work together in one place?”
GLM (Zhipu)searched the webretrieved figma.comnamed figma.comDesigners and developers can collaborate in real-time within Figma, which allows simultaneous editing, commenting, and developer handoff via Dev Mode. Framer and Webflow enable teams to design and publish production-ready websites together on a single canvas. For component-driven workflows, Storybook provides a shared space to build, document, and review UI components in isolation. Miro and FigJam serve as digital whiteboards for brainstorming and strategy alignment. Additionally, GitHub…
“What is the best tool for real-time collaborative design and prototyping?”
ChatGPTsearched the webretrieved figma.comnamed figma.comFigma is the best all-around choice for real-time collaborative design and prototyping. Its multiplayer editing, comments, design systems, and interactive prototypes work smoothly in the browser, making it ideal for distributed product teams. FigJam, included in the ecosystem, adds whiteboarding and workshops. ([figma.com](https://www.figma.com/pricing/?utm_source=openai)) Strong alternatives: - Penpot — best open-source/self-hosted option, with real-time collaboration and prototyping. ([penpot.app](h…
“Where can designers and developers work together in one place?”
ChatGPTsearched the webdid not retrieve figma.comnamed figma.comDesigners and developers can work together in: - Figma — collaborative design, prototyping, comments, and Dev Mode for inspecting specs, assets, and code-oriented handoff. ([help.figma.com](https://help.figma.com/hc/en-us/articles/15023124644247-Guide-to-Dev-Mode?utm_source=openai)) - Penpot — open-source, browser-based design with inspect mode, CSS/HTML output, design tokens, and API access. ([penpot.app](https://penpot.app/collaboration?utm_source=openai)) - Storybook — a shared environment for b…
Same method, three sites
| Site | Named by the engine | AIVS | 95% interval | Measured | |
|---|---|---|---|---|---|
| figma.com | 14 of 16 | 79 | 60–83 | September 2, 2026 | Report |
| allbirds.com | 11 of 16 | 78 | 59–87 | September 2, 2026 | Report |
| notion.so | 5 of 16 | 48 | 38–67 | September 2, 2026 | Report |
The interval is a Wilson 95% interval over eight answers. Asking the same engine the same eight questions three times returned one, two and four mentions, so a single number would claim a precision the measurement does not have.
7,632 sites, read the way a crawler reads them
Beacon fetched 11,782 domains sampled from a Tranco top list with a plain HTTPS request, no browser and no JavaScript, and ran the same ten Citability checks on each. 7,632 answered; the rest timed out, blocked the request, or returned a challenge page and are excluded rather than scored as zero. Your Citability percentile is your position on this distribution.
Weighted towards European markets: .de 22%, .uk 10%, .nl 9%, .fr 7%, .pl 6%. Median 76, mean 72.8.
Citability score, 0–100. Height is the number of sites at that score.
How the score is made
Three layers, kept apart. Two are measured today. The third needs your own traffic data and is shown as planned, with no number, until it is connected.
Blue is an observation. White is our own weighting of observations. Amber is a plan. Blue is never used for anything else.
| Layer | What is recorded | Weight in AIVS |
|---|---|---|
| Visibility | Eight buyer questions, written from your page's own description, asked of a live engine. Per answer: did the engine search, was your domain in what it retrieved, and does the answer name you. Reported with a 95% interval. | 0.6 |
| Citability | Ten checks on the raw HTML, robots.txt and llms.txt: readable without JavaScript, AI answer crawlers allowed, structured data, heading order, quotable passages, attribution, canonical. Deterministic, so no interval; ranked against the corpus. | 0.4 |
| Conversion | Whether AI-referred visits turn into anything. Needs your analytics connected. Nothing is modeled in its place. | planned |
A badge that carries its date
Every scan produces an embeddable badge. It shows the score as measured on that day and links to the full report, so anyone who sees it can check the number. It says “score”, not “verified”: we measured a domain, we did not audit a company.
Below is Beacon's own badge, from Beacon scanning itself. Whatever the score is, that is what shows.
Pricing
Start free. Paid tiers have no self-serve checkout yet; ask and we will set you up.
Free$0One scan a day, no account. | Pro$49/moTeams shipping pages every week. | Agency$199/moAgencies reporting to clients. | EnterpriseCustomYour own scope and terms. | |
|---|---|---|---|---|
| Scans | 1 per day | Unlimited | Unlimited | Volume |
| Visibility and Citability report | yes | yes | yes | yes |
| All ten Citability checks | yes | yes | yes | yes |
| Answer-by-answer engine detail | — | yes | yes | yes |
| History and trend per domain | — | yes | yes | yes |
| White-label reports | — | — | yes | yes |
| Several domains, one account | — | — | yes | yes |
| Direct support | — | — | — | yes |
| Request access | Request access | Request access |