AI Visibility Metrics That Matter

AI visibility metrics are numbers that show how often, and how well, AI answer engines such as ChatGPT, Perplexity and Google AI Mode present your brand to buyers. An answer engine is a tool that writes an answer instead of showing a list of links. You get these numbers by asking the engines the same buyer questions many times and counting what the answers say.

Classic SEO metrics do not cover this. Google's AI features guide says traffic from AI Overviews and AI Mode is counted inside the "Web" search type of Search Console, not on its own line. So your search reports cannot tell you whether an AI answer named you. This page explains which numbers to track. Our parent guide, how to track brand mentions in ChatGPT and Perplexity, shows how to collect them step by step.

Mention rate is the share of answers that name your brand. If 6 of 30 runs name you, your mention rate is 20%. It is the first number most teams look at, and the easiest to misread.

Always split it in two. A branded question contains your name, such as "is [brand] reliable". An unbranded question does not, such as "best [product] for [need]". In Leadee's tracker round of 2026-10-25, the seven Netbase JSC brands were named in 62 of 84 US answers to branded questions (73.8%, 95% interval 63.5-82.0%) but in only 20 of 274 answers to unbranded ones (7.3%, 4.8-11.0%). Unbranded mention rate is the one that shows whether new buyers find you. Our guide on checking if ChatGPT mentions your brand has the full test.

Citation rate is the share of answers that link to a page on your own website. A mention is your name in the text; a citation is a link the buyer can click. Through the official APIs both are data, not guesswork. OpenAI's web search guide returns each citation as a url_citation annotation and notes that "the number of sources is often greater than the number of citations". Perplexity's Sonar docs return a citations list of URLs. Count cited pages, not every page the engine read.

The two numbers can split apart. In the same round, Perplexity named one of the brands in 46 of 252 runs but cited its domain in 41. Web2Print, part of Netbase JSC, was named 7 times in the US and its own domain was cited 0 times, so buyers who clicked landed on other sites. The other way round, CMSmart, part of Netbase JSC, had cmsmart.net as the most-cited domain in its US niche (96 citations, ahead of 89 for woocommerce.com). See the Web2Print case study and our guide to ChatGPT citations.

Position is where your brand appears when the answer is a list: first, third, or not at all. Report the average position only over the answers that name you, and always next to the mention rate. A brand named once in fifth place and a brand named often in second place can have the same average.

Share of voice is your mentions divided by the mentions of every brand in the same question set. It shows how much of the conversation you hold against other market players, not just whether you appear. In the 2026-10-25 round, for the niche of AdminKit (part of Netbase JSC), another market player, MXToolbox, held 48% share of voice with 23 mentions, while AdminKit was named in 0 of 39 unbranded US questions. The AdminKit case study has the full table.

A mention only helps if the answer is right. Check each answer that names you for three things: is it really your company, are the facts correct (price, location, what you sell), and is the tone positive, neutral or negative. The tone check is often called sentiment.

Name mix-ups distort every other metric. On 2026-10-25 Leadee's raw extractor counted 12 "AdminKit" mentions, but most were about an unrelated website template with a similar name, not AdminKit.net. Without a manual check, mention rate and share of voice would both have looked better than they were. If the answers get your facts wrong, follow how to fix wrong information in ChatGPT.

AI answers change from run to run, so every rate is an estimate. A 95% confidence interval is the range the true rate probably sits in, given how many runs you made. Leadee uses the Wilson score interval, described in the NIST/SEMATECH statistics handbook, because it stays sensible with small samples and rates near zero.

Sample size changes what a number can tell you. At the same observed rate of 30%, 3 mentions in 10 runs gives an interval of 10.8% to 60.3%, while 30 in 100 gives 21.9% to 39.6% (arithmetic examples, not measurements). Single runs also disagree: in Leadee's 2026-10-25 data, one run picked at random missed a brand that did appear in 22 of 105 runs (21%). Our comparison of manual and automated tracking has the full count, and our methodology shows how Leadee computes every rate.

AI referral traffic is the visits that reach your site from a link in an AI answer. It matters because it is the step closest to a sale, but it is the hardest to read. The Search Console API returns clicks, impressions, click-through rate and position by query and page, and Google says AI Overviews and AI Mode are counted inside the "Web" search type. You cannot pull the AI share out of that total.

Referral visits from ChatGPT or Perplexity can show up in your analytics as referrals, but only when a buyer clicks. Many read the answer and never visit. So treat traffic as a lagging signal next to mention and citation rate, not as a replacement. The Leadee AI Traffic module puts Search Console, GA4 and AI crawler hits from your own logs next to the tracker numbers, each labelled by source.

Start with the metrics that answer "do new buyers see us, and do they reach us?" and add the rest once you have a stable question set. A short list read every period beats a dashboard nobody trusts.

Leadee is an AI-native platform that stands on the shoulders of giants: it asks ChatGPT, Gemini, Perplexity and Google AI Mode through their official APIs and never scrapes a consumer chat app. The Visibility Tracker reports mention rate, citation rate, position, share of voice and sentiment for your question set, each with a 95% Wilson interval, and tracks other market players with the same method.

Your team owns the question list and decides what to change. To set the metrics up by hand first, follow the parent guide on tracking brand mentions in ChatGPT and Perplexity and copy our prompt tracking sheet. Every other ChatGPT guide is on our ChatGPT SEO hub.