Guide12 min read
How to Measure AI Search Visibility Beyond Rankings
AI search visibility is how often AI answers name your brand, link to your pages as sources and describe you correctly. Rankings and traffic do not show it. Measure five things: mentions, citations, share of voice, answer accuracy and AI referral traffic. Repeat every check, because AI answers change from one run to the next.
Key takeaways
- Five measures cover AI search visibility: mention rate, citation rate, share of voice, answer accuracy and AI referral traffic.
- Repeat each question many times, because one study found less than a 1 in 100 chance of the same list of brands in any two responses.
- Report each AI tool on its own line, because ChatGPT and Gemini showed opposite patterns of naming and linking in one 2026 study.
- Google Search Console now reports impressions in Google's generative AI features once a site has enough impressions, but not ChatGPT, the Gemini app or Perplexity.
- If you run Google Ads, read paid results and AI visibility in one monthly view.
Why do rankings and traffic miss AI search visibility?
Rankings and traffic describe pages and visits. AI answers describe brands and sources. A page can sit in the top ten and still be missing from the answer, and a brand can be named in the answer without anyone clicking. For the wider picture of how AI answers change search, read the future of search with AI answers.
A page can rank without being in the answer
In a March 2026 Ahrefs study of 863K keyword results pages and 4M AI Overview URLs, 37.9% of the URLs cited in AI Overviews also appeared within the first 10 blocks of the results.
An earlier Ahrefs study of 15,000 prompts, August 2025, found that on average only 12% of links cited by ChatGPT, Gemini and Copilot appear in Google's top 10 results for the same prompt. Perplexity was the outlier, at nearly 1 in 3. The methods and dates differ and Ahrefs has reported other figures before, so read these as a direction. A rank report cannot tell you whether you are in the answer.
A citation is not a mention
A citation is a link to your page, and a mention is your brand name in the answer. In a Semrush study with Kevin Indig, June 2026, the team logged 3,981 domain appearances for 115 prompts run across 14 countries and four AI search engines. Almost 62% were ghost citations: the AI used the page as a source link, but the brand name never appeared in the answer.
Most appearances were a source link with no brand name in the answer, and only 13.2% had both.
- Cited, brand not named61.7%
- Named, not cited25.1%
- Both named and cited13.2%
Source: Semrush with Kevin Indig, June 2026, 115 prompts, 14 countries, four AI search engines
Clicks can fall even when a brand appears
In a Pew Research Center study, July 2025, of the browsing of 900 US adults in March 2025, users who saw an AI summary clicked a traditional result link in 8% of visits. Those who did not see one clicked in 15% of visits.
A 2026 field experiment on SSRN by researchers at Carnegie Mellon University and the Indian School of Business, posted as a working paper on SSRN, found that when AI Overviews appeared they reduced outbound organic clicks by 39.8% and increased zero-click searches by 34.5%.
Google says that clicks from results pages with AI Overviews are higher quality, meaning users are more likely to spend more time on the site. The field experiment found engagement on the sites visited was not different across groups. The sources disagree, so check your own analytics.
What should you measure?
Five measures cover AI search visibility. Each needs a fixed set of buyer questions, run on each AI tool, because the numbers come from reading the answers.
| Measure | The question it answers | Where to read it |
|---|---|---|
| Mention rate | How often does the answer name the brand? | The answer text, for each question and tool |
| Citation rate | How often does the answer link to one of your pages? | The sources shown with the answer |
| Share of voice | Of all brands named, what share is yours? | The same runs, counted across the whole question set |
| Answer accuracy | Is what the answer says about you right? | The answer text, checked against your own pages |
| AI referral traffic | How many visits arrive from AI tools? | Your analytics, by referral source |
Source: the definitions are Morsegrid's own.
Mention rate
Mention rate is the share of runs in which the answer text names your brand. Count a mention only when the name is in the words the tool wrote, not in a list of sources. Record it for each question and each tool, because tools can differ a lot, as a later section shows.
Citation rate
Citation rate is the share of runs in which the answer links to one of your pages as a source. Each tool shows sources in its own way. Perplexity says each answer includes numbered citations linking to the original sources. Google says Gemini Apps may provide links to sources and that not all responses include related links or sources, so an empty source list in Gemini is not always a miss.
OpenAI says that search results and citations in ChatGPT can be incomplete, outdated, or incorrect. Open a citation before you count it.
Share of voice
Share of voice is your share of all the brand names in the answers to a fixed question set. Count every brand named across the set, divide your count by the total, and do the same for each competitor.
In a January 2026 study, SparkToro and Gumshoe wrote that measuring percent visibility is probably a reasonable way to see how prominent a brand is within an AI's consideration set. The author is much less likely to read meaning into a brand's position in the list, so count appearances, not list rank.
Answer accuracy
Answer accuracy is whether what the answer says about your brand is correct. Read the text and check it against your own pages: the price, whether the product is in stock, the delivery time, the return window and whether a product still exists. Because results can be incomplete, outdated or incorrect, as OpenAI says above, a named brand with a wrong detail is worth catching.
AI referral traffic
AI referral traffic is the number of visits that reach your site from AI tools. It is the easiest to count and the narrowest view, because it only sees people who clicked. Similarweb, September 2026, reports that AI platforms drove an average of 770.7 million referral visits per month worldwide between June 2025 and May 2026, more than double the year before (+117.4%, desktop and mobile web). That is a total across the web, not a benchmark for your site.
Similarweb, July 2026, also reports that roughly six in ten ChatGPT referred visits land on a homepage rather than a specific page. That share rose from roughly 26 to 29% of referral traffic before ChatGPT's May 2026 update, which replaced footnote-style citations with clickable brand names inside its answers, to about 62 to 63% by late May 2026. Similarweb adds that traffic once filed as "direct, no clear source" may increasingly be AI-driven discovery that does not carry a clean referral tag.
How do you measure answers that change every time?
One check of an AI tool is a sample, not a measurement, so the method has to expect change.
Run each question many times
In a January 2026 study by SparkToro and Gumshoe, 600 volunteers ran 12 different prompts through each of 3 tools a combined 2,961 times. The researchers found a less than 1 in 100 chance that ChatGPT or Google's AI, if asked 100 times, would give the same list of brands in any two responses. SparkToro notes on the page that a co-researcher works at an AI tracking startup, so read the findings with that in mind.
The author's advice is to ask over and over, usually at least 60 to 100 times. Few teams can do that by hand each month, so run your most important questions several times, read the result as a range and keep the wording fixed. Our own AI Search check follows the same logic: answers can change from one check to the next, so one check proves little, and a monthly record shows the direction.
Fix the question set and write it the way buyers ask
Keep the wording the same from month to month, or a change in the result may only be a change in the question. Write long, specific questions. Google's CEO said on the Q3 2026 earnings call that queries in AI Mode are on average three times longer than traditional searches. Add the follow-ups a buyer would ask next, because Google says AI Overviews and AI Mode may use a query fan-out technique, issuing multiple related searches across subtopics and data sources.
Report each tool on its own line
Report each AI tool separately, because the tools behave differently. In the Semrush and Kevin Indig study, June 2026, Gemini mentioned brands in 83.7% of appearances but generated a citation link 21.4% of the time. ChatGPT did the opposite: it cited brands 87% of the time and mentioned them in only 20.7% of answers. A single blended score would hide that gap.
Gemini named brands far more often than it linked to them, and ChatGPT linked far more often than it named them.
- Gemini names the brand83.7%
- Gemini links the page21.4%
- ChatGPT links the page87%
- ChatGPT names the brand20.7%
Source: Semrush with Kevin Indig, June 2026, 3,981 domain appearances
What can Search Console and analytics tell you?
Search Console and your analytics each cover one slice. Neither shows mentions inside an answer, so they support the five measures and do not replace them.
Search Console's generative AI insights
Google's Search Console Help says that as of August 31, 2026, it has rolled out these insights to all websites worldwide. It defines impressions as how many times links to your site were shown to a user in a generative AI feature on Google Search. The report may not show when a site has too few generative AI impressions.
It counts impressions, not mentions in the answer text, and it covers Google Search only, so ChatGPT, the Gemini app and Perplexity do not appear in it.
Your own analytics
Create one segment in your analytics for AI sources, listing the referral domain of each AI tool you care about. Track homepage landings and specific page landings separately.
Watch direct traffic too, but do not assign a rise in it to AI without other evidence.
How does AI visibility connect to paid search and predictable ROI?
This section reports Google's own advice to advertisers, not a Morsegrid service, and it matters only if you run Google Ads.
Ads are moving into AI answers. In a September 2026 Think with Google article by Google's VP of Global Ads, Google says ads are eligible to show above or below AI Overviews in 200+ markets and that it is actively testing ads in AI Mode.
Bidding has changed too. Google Ads Help says that starting on 17 August 2026, Google updated its bidding systems to deliver more consistent and predictable performance based on the targets you set, even when you make budget adjustments. In a November 2026 Think with Google interview with Google's Head of Search+ for EMEA, the line is that your target controls your efficiency, while your budget simply dictates the total revenue you capture at that return. Google also says to work directly with your finance team to set Target ROAS or Target CPA from true unit economics, such as gross margin, net profit contribution or customer lifetime value, because an artificially high target means you walk away from profitable auctions.
What this means for measurement
If you run Google Ads, read paid results and AI visibility in one monthly view. A target set from real unit economics gives you a steadier paid baseline, which in our reading makes movement elsewhere easier to see. No source we read says AI visibility lifts paid returns, so we do not claim that it does.
A monthly routine you can run yourself
You can run this routine each month with a spreadsheet. Five steps, in order:
- Write 20 buyer questions in your customers' words, taken from support tickets, reviews and sales calls.
- Run each question on ChatGPT, Google AI Overviews, Google AI Mode, Gemini and Perplexity, and repeat the most important ones.
- Record whether the brand was named, whether a page was linked, which competitors appeared and whether the text was accurate.
- Compare the result with your Search Console impressions and your AI referral visits.
- Pick two fixes, make them, and run the same questions next month. The guide to getting cited in ChatGPT and AI search covers what each vendor says helps.
If you would rather not run it by hand, our AI Search work covers whether you are named, where you appear, who is named instead and which sources each answer cites each month across ChatGPT, Google AI Overviews, Google AI Mode, Gemini and Perplexity. You get one monthly report and one plan, and the first check is free. You can see how we work together too.
Frequently asked questions
- What is AI search visibility?
- AI search visibility is how often AI answers name your brand, link to your pages as sources and describe you correctly, across tools such as ChatGPT, Google AI Overviews, Google AI Mode, Gemini and Perplexity. A citation (a source link) and a mention (your name in the answer text) are different results.
- Can I trust a single check of ChatGPT?
- No. A SparkToro and Gumshoe study, January 2026 found a less than 1 in 100 chance that ChatGPT or Google's AI gave the same list of brands in any two responses, across 2,961 runs. Treat one check as a sample, repeat your important questions and compare month to month.
- Can Search Console show my visibility in Google's AI features?
- Partly. Google Search Console Help says that as of 31 August 2026 it has rolled out generative AI insights to all websites, counting how many times links to your site were shown in a generative AI feature on Google Search. It covers Google only and counts impressions, not mentions in the answer.
- Is AI referral traffic big enough to matter?
- It depends on your site, so check your own analytics first. Similarweb, September 2026, reports that AI platforms drove an average of 770.7 million referral visits per month worldwide between June 2025 and May 2026, but that is a total across the web, not your share. Referral visits also miss people who saw your brand and did not click.
Start with 20 buyer questions across five AI tools and write down what you see. If you want the first check run for you, book the free AI visibility check on the AI Search page.
Keep reading
Guide
The Future of Search With AI Answers: What Changes in 2027
AI answers now sit above the links and fewer clicks reach websites. See what changed, what did not, and how a brand prepares for 2027.
16 min read
Guide
How to Get Your Brand Cited in ChatGPT and AI Search
How ChatGPT, Google AI Overviews, AI Mode, Gemini and Perplexity find and cite sources, and what a brand can do to be named in each one.
13 min read