Why We Started Tracking AI Mentions
For twenty years, the question was: "Where does your business rank on Google?" That question still matters. But a parallel question is emerging: "Does your business appear when someone asks an AI for a recommendation?"
When a potential customer asks ChatGPT "What's the best dentist in Edmonton?" or asks Perplexity "Who does commercial paving in Calgary?" or gets a Google AI Overview for "optometrist near me," the AI generates an answer. That answer may or may not include your business. If it does, that's a discovery channel. If it doesn't, that's information you need.
We started tracking AI mentions because we track everything else. We monitor 60,000+ data points daily across Google Ads, organic search, page performance, and competitor activity. AI visibility is a new data stream, and ignoring it would leave a gap in our understanding of where our clients' businesses appear online.
The tracking isn't driven by hype. Nobody's organic traffic from AI referrals has replaced Google yet. For most local businesses, AI-generated traffic is a rounding error compared to traditional search. We track it because the trajectory matters more than the current volume, and because having 12 months of baseline data when the optimization playbook matures will be significantly more valuable than starting from zero.
The 10 Platforms We Monitor
Each AI platform discovers and cites businesses differently. Understanding the mechanics of each one is necessary to interpret the data.
Google AI Overviews. Google's AI-generated summaries appear at the top of search results for an increasing percentage of queries. When someone searches "best paving company in Edmonton," Google may generate an AI Overview that names specific businesses. This is the highest-impact AI visibility channel because it appears within the Google search experience itself, where the traffic already is. Visibility here correlates heavily with existing organic ranking signals.
ChatGPT. OpenAI's conversational AI has a web browsing capability and increasingly includes business recommendations in its responses. When a user asks ChatGPT for local business recommendations, it synthesizes information from multiple sources. The model's knowledge has a training data cutoff, but web browsing fills some gaps. ChatGPT is the highest-traffic AI platform globally, but its local business recommendation capability is still developing.
Perplexity. Perplexity functions as an AI-powered search engine that cites its sources. When it recommends a business, it links to the source material. This makes Perplexity mentions particularly trackable because the citation includes a URL. The referral traffic is measurable in GA4. Perplexity's audience skews toward researchers and professionals who want cited answers rather than generated text.
Claude. Anthropic's AI assistant handles business-related queries with a research-oriented approach. Claude's training data includes business directories, review sites, and industry publications. For local business queries, Claude tends to recommend based on published reviews and directory presence rather than generating novel recommendations.
Gemini. Google's AI assistant has deep integration with Google's own data, including Google Maps and Google Business Profile information. Gemini recommendations for local businesses draw heavily from GBP data, making GBP optimization directly relevant to Gemini visibility.
Copilot. Microsoft's AI assistant, integrated into Bing and Microsoft 365, draws from Bing's search index. Businesses with strong Bing presence see Copilot mentions. Given that Bing's market share is smaller than Google's, the volume is proportionally lower, but the audience includes a significant enterprise user base.
You.com. An AI search engine that provides cited answers. Lower traffic than the major platforms, but its search-engine format means mentions are structured and attributable.
Phind. An AI search engine focused on technical and professional queries. Relevant for B2B services and technology companies rather than consumer-facing businesses.
Meta AI. Meta's AI assistant is integrated into WhatsApp, Instagram, and Facebook Messenger. When users ask Meta AI for local recommendations, it draws from Meta's business directory and Facebook page data. The integration with messaging platforms means these recommendations happen in private conversations, making them harder to track but potentially high-impact for conversion.
AI-enhanced search results. Beyond dedicated AI platforms, traditional search results increasingly include AI-generated elements: featured snippets that synthesize multiple sources, "People also ask" boxes with AI-generated answers, and knowledge panels with AI-curated information.
| Platform | Primary Data Source | Trackable Via | Current Volume |
|---|---|---|---|
| Google AI Overviews | Google Search index | SERP monitoring, DataForSEO | Growing, tied to search volume |
| ChatGPT | Web browsing + training data | LLM Mentions API | Low for local, growing |
| Perplexity | Web search, cited sources | GA4 referral traffic, LLM API | Small but attributable |
| Claude | Training data, publications | LLM Mentions API | Small |
| Gemini | Google Maps, GBP, Search | LLM Mentions API | Moderate, Google-integrated |
| Copilot | Bing index | LLM Mentions API | Small, enterprise skew |
| You.com | Web search, cited | GA4 referral traffic | Very small |
| Phind | Web search, technical focus | GA4 referral traffic | Very small, B2B only |
| Meta AI | Facebook/IG data | Limited, platform-internal | Unknown, hard to measure |
How We Track It
We use two complementary approaches to measure AI visibility. Neither is perfect, but together they provide a useful picture.
DataForSEO LLM Mentions API. This API queries AI platforms with specific prompts and returns whether a business is mentioned in the response. We run client-relevant queries (service + location combinations) across the platforms and record whether the client's business appears, how prominently, and what competitors appear alongside it. This gives us a snapshot of AI visibility at a specific point in time for specific queries.
The API approach has limitations. AI responses are non-deterministic: the same query can produce different answers on different days. We address this by running queries repeatedly and tracking mention frequency rather than treating any single response as definitive. A business that appears in 7 out of 10 queries for a term has meaningfully different visibility than one that appears in 1 out of 10.
GA4 referral traffic filtering. When someone clicks a link in an AI platform's response and visits a client's website, that visit appears in GA4 as a referral from the AI platform's domain. We filter GA4 referral data for known AI platform domains: chat.openai.com, perplexity.ai, you.com, and others. This gives us actual traffic volume from AI sources.
The referral approach also has limitations. Not all AI interactions result in a website click. A user who gets a business recommendation from ChatGPT might call the business directly without visiting the website. The referral data captures only the subset of AI visibility that produces a measurable website visit.
Between the two approaches, we get both visibility (are you being mentioned?) and traffic (are those mentions driving visits?). The visibility data is leading: it shows where you appear before that appearance translates to traffic. The traffic data is confirming: it validates that visibility is producing actual engagement.
What We've Actually Found
We want to be direct about the current state of the data. For most local businesses, AI-generated traffic is small. Most of our clients see fewer than 50 AI-referral visits per month. Some see fewer than 10. This is not a channel that's replacing Google organic traffic or Google Ads conversions in 2026.
What we have found, across the clients where we have enough data to draw conclusions:
Structured data correlates with AI visibility. Clients whose websites have comprehensive schema markup (LocalBusiness, Service, FAQ, Review) appear in AI responses more frequently than clients without structured data. This isn't a causal claim; businesses with well-built websites tend to have both good schema and good content, and AI platforms draw from both. But the correlation is consistent enough that we consider schema markup a prerequisite for AI visibility, not an optional enhancement.
Google Business Profile completeness matters for Gemini and Google AI Overviews. Clients with fully completed GBP profiles (all services listed, Q&A populated, regular posts, extensive photo galleries) appear more frequently in Gemini responses and Google AI Overviews than clients with minimal profiles. This makes sense given that Gemini draws directly from Google's data ecosystem.
llms.txt files are appearing in citation contexts. Several clients where we've implemented llms.txt files (machine-readable summaries of business information designed for AI consumption) are seeing their llms.txt content reflected in AI responses. The sample size is small, but the pattern is that AI platforms with web browsing capabilities find and use this structured business summary when formulating responses.
Review volume influences recommendations. AI platforms frequently cite review volume and rating when recommending local businesses. A business with 200+ Google reviews and a 4.7 rating appears in AI recommendations more often than a competitor with 30 reviews and a 4.5 rating. The AI platforms appear to use review signals as a proxy for business quality, similar to how Google's organic algorithm uses them.
Content depth matters more than content volume. A dental practice with one comprehensive page about dental implants (2,000+ words covering cost, procedure, recovery, candidacy) appears in AI responses for implant-related queries more often than competitors with five thin pages covering the same topics superficially. AI platforms seem to favor authoritative, comprehensive content that can be directly quoted or synthesized.
The Honest Framing
We can't optimize your way into ChatGPT's answers yet. Nobody can, and anyone who claims they can is selling something that doesn't exist.
AI platforms don't have a transparent ranking algorithm like Google (which itself isn't fully transparent). There's no equivalent of "title tags" or "backlinks" that reliably influence AI responses. The platforms are evolving their recommendation mechanics rapidly, and what works today may not work next month.
What we can do:
Tell you whether you're showing up. If your business appears in AI responses for your key service queries, you know. If it doesn't, you know that too. This visibility data is the starting point for any future optimization work.
Track changes over time. When we add structured data to your website, does your AI visibility change in the following months? When we publish a comprehensive service page, does it start getting cited? The baseline data lets us measure the impact of changes that might otherwise be invisible.
Monitor your competitors. When an AI platform recommends a competitor for a query that should include your business, that's competitive intelligence. We can see which competitors appear and for which queries, which tells us something about what the AI platforms consider authoritative in your market.
Prepare for the shift. AI-generated answers are eating into traditional search clicks. Google AI Overviews appear on an increasing percentage of queries. ChatGPT and Perplexity are growing in usage. The businesses that understand their AI visibility profile now will be better positioned to optimize when the playbook matures than businesses that start measuring from zero.
The trajectory is clear even if the timeline isn't. Five years from now, "where does your business appear in AI answers?" will be as standard a question as "where does your website rank on Google?" is today. The businesses that have been tracking this data for years will have a compounding advantage over those that start tracking when it becomes obvious.
What We're Building Toward
Our AI visibility monitoring is integrated into the same infrastructure that monitors everything else. The data flows into the same centralized database. The same analysts review AI visibility alongside organic rankings, Google Ads performance, and site health. This integration matters because AI visibility doesn't exist in isolation: it's influenced by the same signals (content quality, structured data, online reputation) that drive traditional search performance.
When we build a comprehensive service page optimized for organic search, we're also building content that AI platforms can cite. When we optimize a Google Business Profile for local pack visibility, we're also building the data that Gemini uses for local recommendations. When we implement schema markup for rich results in Google, we're also providing structured data that AI platforms can parse.
The optimization work isn't separate. It's the same work producing value across an expanding set of discovery channels. AI visibility is an additional return on existing SEO investment, not a separate line item.
We don't know yet which AI platforms will matter most in two years. We don't know which optimization tactics will be most effective. What we do know is that having baseline data across all the platforms, collected consistently over time, will be the foundation for whatever optimization approach emerges. And that foundation is being built now, for every client, as part of the monitoring infrastructure we already run.
Learn more about our SEO services and the infrastructure behind our monitoring in How We Monitor 60,000 Data Points a Day. For more on how we build content that serves both traditional and AI-driven discovery, see our approach.