Measure AI visibility with four sources, not one: Google Search Console for AI Overview impressions, Bing Webmaster Tools for Copilot citations, a weekly prompt check for ChatGPT and Gemini, and GA4 for the visits AI actually sends. Each one sees a different slice, and only together do they tell you where you stand. All four are free to start, and we run every one of them on this site each week.
Why is AI visibility hard to measure?
AI visibility is hard to measure because no single tool sees every engine, and most report exposure without clicks. Google counts AI Overview impressions but not clicks from them. Bing counts citations but samples the data. ChatGPT reports nothing to you at all. So you assemble the picture from several partial views and track the trend, not one number.
The other trap is reading clicks per keyword the old way. An AI answer fans one question out into several sub-questions and cites a different page for each. A page can be quoted all week and get few clicks, and a keyword report will call that a failure.
Where do you see your brand in Google's AI answers?
Open Search Console, go to Performance, then the Generative AI report. It counts how often your pages appeared inside AI Overviews and AI Mode. You get impressions only: no clicks, position or queries, and they already sit inside your web totals, so never add the two. Divide a page's AI impressions by its web impressions to get its AI share.
Our numbers from August 3 to November 2: 9,190 AI impressions across 76 pages. The weekly count went from 184 in early August to 1,971 in the last week of October. Our most-shown page had 4,046 AI impressions on 14,860 web impressions, a 27% share. One comparison page reached 45%. The Search Console API does not expose this report yet, so we read it in the browser every Monday.
How do you track citations in Bing and Copilot?
Bing Webmaster Tools has an AI Performance report (in beta) with a Citations and Pages view. It shows how often Copilot and Bing's AI partners cited each of your pages, and the grounding queries behind them. Treat it as directional, because Bing calls the data a sample, and it is the clearest view of which page got quoted for which query.
We had 2,200 Bing AI citations from August to early November, about 5 cited pages a day, across 54 pages. One guide took 35% of them. The same report lists opportunity queries where other sites get cited and you do not. That list is how this post got written: "measuring brand visibility ai" was on it.
How do you check ChatGPT, Gemini and Perplexity?
Run a fixed set of prompts on a schedule and record who gets cited. Pick 20 to 30 questions your buyers ask, run each on every engine you care about, and note whether your domain is cited, mentioned by name, or absent, plus which competitors show up instead. Keep the prompts fixed, so week-over-week changes mean something.
You can do this by hand in a spreadsheet, or with a prompt tracker that runs the set for you. We track 25 prompts across ChatGPT, Gemini and Google AI Overviews, 75 answers per run. Our November 3 run: 3 of 75 answers cited the site and 2 more named it. A competitor was named in 16 answers and cited in none. Being named and being cited are different wins, so log both.
How much traffic does AI actually send?
Check GA4's traffic acquisition report by session source and medium. Look for chatgpt.com, perplexity, gemini and copilot rows next to google / organic. AI referrals are still small for most sites, so this number keeps you honest: visibility that sends no one is worth knowing, but it is not yet a channel.
From October 4 to November 2 we had 333 sessions from Google organic search and 7 from AI assistants, about 48 to 1. Search is still the main channel for us. We watch AI referrals; we do not chase them at the cost of search.
What should a weekly AI visibility report include?
Keep it to one line per source, each compared with the week before: AI impressions and AI share for your key pages, Bing citations by page, prompt-check results (cited, named, absent), and AI referral sessions next to organic. Add one line on branded search, because people who meet you in an AI answer often search your name next.
- Google AI impressions: weekly total, plus AI share on your five most important pages.
- Bing citations: total, top cited pages, and new opportunity queries.
- Prompt check: cited, named and absent counts, and which competitors replaced you.
- AI referrals: sessions and conversions from AI sources against organic search.
- Branded search: impressions and clicks on queries that contain your name.
What do you do with the numbers?
Act on gaps, not totals. A page with high web impressions and a low AI share needs a short answer an engine can quote: a 40 to 60 word reply under a question-shaped heading. A query where competitors get cited and you do not needs a page that answers it. A cited page that sends no visits needs a clearer next step on the page.
The pages engines quote most on our site share a pattern: an answer in the first paragraph, question headings with short answers, a comparison table where one fits, and an FAQ block. If you want the longer version of that playbook, read how to get recommended by AI. For the content that engines read about your brand in the first place, start with how to build an AI-ready brand.
Where should you start?
Start with a baseline of your own site. The brand snapshot audit scores any website on positioning, proof, voice and AI readability in about a minute, and shows where it is weakest. Then turn on the Search Console and Bing reports, write your first 20 prompts, and take your first reading this week.
If the gap is that AI tools do not know your brand well enough to describe it, the fix is a knowledge base they can read. The professional knowledge bases course walks through building one.
AI answers repeat what they can find about you. In Brand Architect, Ophelia builds your knowledge base from your website or materials and keeps it current as the market moves. Build your knowledge base →