AI tracking companies keep promising to show which brands win in conversational search. Rand Fishkin says the trust is gone. The main issue is simple. No one can check if a specific AI prompt actually led to a brand mention or a real sale. There is no way to audit these claims. This isn’t a small glitch. It cuts to the core of how digital media measures influence and return on investment.
HubSpot's AI Search Sensor, launched to monitor how ChatGPT, Gemini, and Perplexity cite brands, explicitly provides only a high-level overview and does not reveal precise brand visibility metrics.
The data sources behind these tools are just as unclear. Most depend on clickstream panels. These only catch the first prompt. They never see what happens inside the app or through API calls. Companies selling AI tracking almost never say who supplies their data. Fishkin even questions if some use real clickstream data at all. Some may just say they do for marketing.
Personalization makes things messier. Every past chat with an AI model changes the next answer. Controlled tests are impossible. Comparing results across users doesn’t work. The idea that one “visibility percentage” can sum up brand presence in AI answers ignores how much personalization changes everything.
Meltwater's GenAI Lens focuses on identifying where brands are mentioned in AI-generated answers, which sources are cited, and how competitors are compared-reflecting an industry trend toward monitoring citations and mentions rather than direct sales impact.
Fishkin doubts the link between AI brand mentions and real sales. He points out that even old-school metrics like “where did you hear about us?” or reported jumps in “visibility percentage” rarely match up with direct traffic or revenue. Marketers demand hard proof from PR, events, or SEO. Why settle for less from AI analytics?
Fishkin’s advice is blunt. Invest in brand mentions. Track their growth. Don’t read too much into the rest. This matches findings from a previous investigation. Concrete signals-like recent publication dates and clear item counts-still drive results in Google and AI search. Speculative metrics mislead.
For publishers and content creators, the warning is direct. Until AI tracking tools show their data sources, open up their methods, and prove real impact, treat their numbers with care. The industry’s rush to measure AI-driven influence risks repeating old analytics mistakes. Flashy numbers don’t guarantee real insight. The only safe move is to focus on what can be checked and demand the same standards from AI analytics as from every other channel.
An official HubSpot product description confirms it. Even top vendors split high-level monitoring from precise brand visibility measurement in AI answers. Detailed metrics are only for paid plans.
