Once a business starts optimising for AI search — structuring content to be quoted, building the trust signals that get it cited — the very next question is the right one: is it actually working? With classic SEO you can check your Google rankings and watch your organic traffic. With AI search, visibility is murkier. An AI assistant might recommend you to thousands of people who never appear in any traffic report, because they read the answer and didn't click. So how do you know?
You can't manage what you can't see, and AI visibility genuinely is harder to see than a ranking. But it's not invisible. Here's how to measure whether AI answers are mentioning your business and sending you customers.
01Why AI visibility is harder to measure
Classic search gives you a rank and a click. AI search breaks both. First, much of its value is a mention without a click — the assistant names you, the user is influenced, but no visit is recorded, so a big chunk of the benefit is invisible to traffic tools. Second, AI answers are variable — ask the same question twice and the response can differ, and it changes as the underlying models update, so there's no stable "position #3" to check. And third, the tooling is young: the mature rank-tracking industry that grew up around Google doesn't yet have a settled equivalent for AI answers. None of this makes measurement impossible — it just means combining methods rather than trusting one number.
02Three ways to actually measure it
1. Test directly — ask the questions your customers ask
The most direct method costs nothing: open the major AI engines (Google's AI Mode and AI Overviews, ChatGPT, Gemini, Perplexity) and ask them the real questions your customers ask — the ones where you'd want to be recommended. Note whether you're mentioned, how you're described, and who's mentioned instead of you.
Do this systematically: keep a list of your priority questions, check them across engines, and record the results so you can see change over time. It's manual, but it tells you exactly what a potential customer sees when they ask — which is the thing that actually matters.
2. Watch your referral traffic
When someone does click through from an AI platform, it shows up in your analytics as referral traffic from that source. Watch for visits coming from AI engines — and check whether that traffic is growing over time. It won't capture the no-click mentions, but a rising trend of AI-referred visits is real evidence that you're being surfaced, and it's traffic that tends to be high-intent, since the person has already been pointed to you as an answer.
Set this up properly so AI referrals are identified rather than lost in a generic bucket — the kind of tracking our conversion and analytics team puts in place.
3. Monitor branded and question-shaped search behaviour
Supporting signals help complete the picture: are more people searching for your brand name directly (a sign that awareness — possibly from AI mentions — is growing)? Is your content earning the featured snippets and clear-answer positions that feed AI answers, visible in Search Console? These don't isolate AI's effect perfectly, but together with direct testing and referral data they show whether your visibility is trending the right way.
03What to track over time
Because AI answers shift, the useful practice is periodic monitoring, not a one-off check. A simple, sustainable routine:
- Your priority questions: are you mentioned across the major engines, and is that improving? (Direct testing.)
- AI referral traffic: is it present and growing? (Analytics.)
- How you're described: when you are mentioned, is it accurate and favourable?
- The gaps: which important questions don't surface you, and who surfaces instead?
Review on a regular cadence, and treat the gaps as your to-do list — the questions where you're absent are exactly where more of the optimisation work should point next.
04Frequently asked questions
How do I know if my business is mentioned in AI search?
Ask the major AI engines — Google AI Mode and AI Overviews, ChatGPT, Gemini, Perplexity — the questions your customers would ask, and note whether you're mentioned and how. Doing this systematically across your priority questions, and repeating it over time, is the most direct way to see what potential customers actually get when they ask.
Can I see traffic from ChatGPT and AI search in my analytics?
Yes, when someone clicks through from an AI platform it appears as referral traffic from that source in your analytics. Set tracking up so AI referrals are identified rather than lumped into a generic bucket. Note that this only captures clicks — the mentions where someone reads the answer without clicking won't show up, so pair it with direct testing.
Is there a tool to track AI search visibility?
The tooling is emerging and less mature than traditional rank trackers, and no single tool yet captures AI visibility perfectly. The reliable approach today combines direct testing (asking the engines your key questions) with referral-traffic analytics and supporting signals like branded search — rather than relying on one dashboard.
How often should I check my AI search visibility?
Regularly rather than once, because AI answers change as models update and content shifts. A periodic routine — checking your priority questions across engines and reviewing AI referral traffic on a set cadence — shows the trend, which matters far more than any single snapshot that might read differently the next day.
05Where this leads
Optimising for AI search without measuring it is flying blind. The measurement is messier than checking a Google ranking, but between direct testing and referral data you can build a genuine picture of whether AI answers are mentioning you, how, and whether it's improving — and turn the gaps into your next priorities.
This closes the loop on optimising for AI search and appearing in AI Overviews. If you'd like this set up and monitored for you, our conversion and analytics and SEO teams do exactly that. Ask about AI-search tracking.
