AI visibility tracking: measure the visits and revenue, not just the mentions

AI visibility tracking measures how your brand shows up in ChatGPT, Perplexity, Claude, Gemini and Google's AI answers. Most tools do it by asking those assistants a fixed list of prompts every day and recording whether you're mentioned, cited or recommended, and how you compare with competitors.

That's useful, and it's half the picture. A mention in an answer is an impression. What a founder needs to know is what happens next: did anyone click, did they sign up, did they pay? Prompt trackers can't see that, because it happens on your site. This guide covers the whole chain and what measures each link, including what Lantern adds at the end: AI as a channel, AI crawlers by page, and revenue from AI-referred customers.

The AI visibility chain, from crawl to customer

Visibility in AI search runs in five steps. Each one needs a different measurement:

Prompt trackers own step 3. Your analytics should own steps 1, 2, 4 and 5. Most sites measure step 3 and nothing else, which is like measuring ad impressions and never looking at sales.

What prompt-tracking tools do well, and what they miss

Tools in this category, such as Profound, Peec AI, Otterly.AI, Semrush's AI Visibility Toolkit and Ahrefs Brand Radar, run prompts against AI assistants and report on the answers. They're good at:

• Share of voice: how often you're named against competitors for the prompts you care about.
• Citations: which of your pages, and which third-party pages, the assistants cite.
• Sentiment and accuracy: whether the answer describes you correctly.
• Trends: whether a content change moved you up for a set of prompts.

Their limits come from the method:

• The prompts are a sample. Real people phrase questions in endless ways, add context about their situation, and have memory turned on. A tracked prompt is an approximation of what your buyers ask.
• A mention isn't a visit. Being named fifth in a list of ten may send no one.
• They stop at the answer. They can't tell you whether AI-referred visitors convert better or worse than Google's, or which assistant brings customers who pay.

Steps 1 and 2: are the assistants reading you?

Before an assistant can recommend you, its crawlers have to read your pages, and before it cites you in a live answer, its fetcher has to retrieve them. Neither shows up in Google Analytics, because crawlers don't run JavaScript. (Details in why GPTBot doesn't show up in GA4.)

Lantern tracks about 35 crawlers from a few lines on your server, grouped by the company that runs them and by what they were doing: training (GPTBot, ClaudeBot), AI search (OAI-SearchBot, PerplexityBot) and assistant fetches (ChatGPT-User, Claude-User). The assistant fetches matter most: each one means a real person just asked an assistant a question your page might answer.

What to look for

• Are your pricing, product and comparison pages being read, or only the homepage?
• Did a robots.txt rule or a firewall setting block a crawler you wanted?
• After you publish a page, how long until the AI crawlers read it?
• Which pages get fetched for live answers? Those are the questions people are asking about you.

Step 4: the visits AI sends you

Most analytics tools file chatgpt.com, perplexity.ai and claude.ai under Referral, next to every blog that ever linked to you. Worse, many visits from AI apps arrive with no referrer and land in Direct. Lantern gives AI assistants their own channel and shows each assistant separately, so you can see:

• How many visitors each assistant sends, and the trend week by week.
• Which pages they land on. If it's your pricing page, they arrive ready to compare.
• How they behave: bounce rate, pages per visit, time on site, against Google visitors.

For the visits that arrive without a referrer, ask at sign-up. A "How did you hear about us?" question with an AI assistant option catches people who read about you in ChatGPT and typed your URL later. (See the how-did-you-hear-about-us survey guide.)

Step 5: the customers and revenue AI brings

This is the step that justifies everything else, and the one almost nobody measures. Connect your payment provider to Lantern (Stripe, Paddle, Polar, Lemon Squeezy, Razorpay, Cashfree or Dodo Payments) and every payment, renewal and refund is credited to the channel that first brought the customer. On many founders' sites, ChatGPT already sits near the top of that list, often with a higher average order than search.

Once you know that ChatGPT-referred customers pay, say, $57 a month against $49 from Google, a prompt tracker's share-of-voice chart becomes a revenue forecast instead of a vanity metric.

A practical AI visibility stack for a small team

1. Analytics that sees AI (Lantern): crawler reads, AI visits by assistant, conversions and revenue. This is the foundation, because it measures outcomes.
2. A sign-up survey with an AI assistant option, recorded on the sign-up goal so you can compare it with the tracked channel.
3. Optionally, a prompt tracker once AI is a meaningful channel, to see which prompts you win and lose against competitors.
4. A monthly review: which pages AI reads, which it sends people to, and which of those visitors pay. Then improve the pages that sit on that path.

If you've connected Lantern to Claude, the monthly review can be one prompt: "Compare AI visitors, sign-ups and revenue this month against last, by assistant, and list the pages AI crawlers read most." (Setup: connecting your analytics to Claude.)

How to improve AI visibility, using what you measure

• Make sure the crawlers can read you. Check robots.txt and bot rules, render content in HTML rather than only in JavaScript, and publish an llms.txt that points to your key pages.
• Strengthen the pages AI already sends people to. If ChatGPT visitors land on pricing, make pricing answer the questions they come with.
• Write the comparison pages assistants cite. "X vs Y" and "best tools for Z" pages are what assistants retrieve when people ask for recommendations.
• Get mentioned where assistants read. Reddit threads, reviews and independent articles are frequent sources for AI answers.
• Re-measure. Watch crawler reads, AI visits and AI revenue after each change. If they don't move, the mention didn't matter.

Questions people ask

What is AI visibility?

How often, and how favourably, AI assistants such as ChatGPT, Perplexity, Claude and Gemini mention, cite or recommend your brand when people ask them questions.

Can Google Analytics track AI visibility?

Only partly. GA4 records some visits from AI assistants as referrals, misses the ones without a referrer, and can't see AI crawlers at all. See how to track ChatGPT traffic for the details.

Do I need a prompt-tracking tool?

Not to start. First measure whether AI sends you visitors and customers; that's cheaper and tells you whether the channel matters for you. Add prompt tracking when it does, to understand why you win or lose.

Does Lantern track brand mentions in ChatGPT answers?

No. Lantern measures what happens on your site: AI crawlers reading your pages, visitors arriving from assistants, and the sign-ups and revenue they bring. It's the outcome side of AI visibility, and it pairs well with a prompt tracker.

How much does Lantern cost?

Plans start at $8 a month after a 7-day free trial with no card. The AI channel, AI crawler tracking and revenue attribution are on every plan.