Who Gemini names when you ask for an AI recommendation tracker.

What we asked, and how.

outrankfast tracks buying questions for its own site the same way it tracks them for customers. By an AI recommendation we mean an assistant naming a business in its answer, which is what a buyer sees. From 17 to 26 September 2026 we asked Gemini, grounded with Google Search, four questions once a day:

• How do I track AI recommendations for my business?
• I'm a marketer and I want to track AI recommendations. What are the best tools for that?
• I'm a marketing manager and I want to track AI recommendations. What are the best tools for that?
• I'm a business owner and I want to improve my marketing results. What are the best AI recommendation tracking and content generation tools for that?

That produced 42 answers, and Gemini searched the web before all 42. A model read the business names out of 40 of them, and a name was kept only if it appears in the answer word for word. The other two answers were not read for names, so the counts below are out of 40.

outrankfast was named in none of them.

0 of 42. The site had no page answering any of these questions, and Gemini names businesses from the pages its searches turn up. This blog is our attempt to change that. The same four questions keep running daily, so whether it worked will show in the same numbers.

Who Gemini named.

Counting the businesses that sell tools in this category, out of 40 answers:

The assistants themselves came up as the things to track: ChatGPT in 34 answers, Perplexity in 22, Claude in 16, Google AI Overviews in 15 and Gemini in 14.

Three patterns stand out.

• The question decides the list. The plain question, "How do I track AI recommendations for my business?", drew trackers: Mentionova, Profound, Otterly.AI, Peec AI and Semrush's AI toolkit led, and none of its eight answers named a writing tool. The business owner's question, which asked for content generation too, drew writers into 14 of its 15 answers.
• The lists do not settle. No business appeared in all the answers to any single question. The two marketer questions produced 53 different businesses across 17 answers.
• General suites crowd in. HubSpot, Similarweb and several personalization platforms appeared on questions about tracking AI answers, and the pages Gemini cited most were general lists of AI marketing software.

In this category, named means recommended.

Gemini mentioned few businesses in passing. Across the 40 answers it named businesses in this category 284 times, and 273 of those were recommendations: an entry in a list, a heading or bold text. An answer named about seven such businesses on average.

So in this category the useful question is whether you are on the list at all. A business left off the list does not turn up elsewhere in the answer: only 11 of the 284 mentions came in passing.

What it searched for.

Gemini ran 80 different searches across the 42 answers, a little over three per answer. The ten it ran most:

Most are the question compressed into a keyword string, often with "business marketing" added. Those strings are what a page has to answer to be found.

Gemini read the question two ways.

In 16 of the 42 answers, at least one search treated "AI recommendations" as recommendation engines or personalization software: "ai recommendation engines for marketing", "ai tools for personalized marketing recommendations", "ai recommendation system performance monitoring". It ran 14 different searches of that kind.

Those answers cited pages about a different product category, such as Teradata's guide to AI recommendation engines and Insider One's list of AI personalization tools, each in 5 answers. For anyone writing about this category, the fix is to put a plainer term next to the ambiguous one: AI visibility, or whether ChatGPT names you.

Which pages it cited.

The pages cited in the most answers, out of 42:

All eight pages cited in 10 or more answers are lists of tools. The how-to question went elsewhere: its most cited pages were guides with titles close to the question, such as Omni Online Strategies' how to track whether AI tools are recommending your business, cited in 8 answers.

Each answer drew on about ten sites.

The 42 answers cited between 5 and 18 different websites each, about 10 on average. A few sites carried a large share. Zapier was cited in 23 answers, through two different lists. Marketer Milk was cited in 21, One Little Web in 16 and Rivo in 15.

Vendors in the category appear among the sources too. Nightwatch's own blog was cited in 9 answers and Semrush's in 8, and both businesses appear in the table of names above. A vendor that publishes a clear list of tools in its own category gives the assistant a page to read and a name to find on it.

The limits of this sample.

This is one assistant, four questions and ten days. Gemini answered through Google's API, grounded with Google Search, which means no personal history shaped the answers; the Gemini app can answer the same question differently for a signed-in user. The questions are in English and cover one category.

Read the counts as a snapshot of this category in late September 2026. The same questions keep running daily, and the counts will move as pages are published and models change.

What this means for your category.

Gemini names the businesses that appear on the pages it reads, and for a "best tools" question it reads lists. Two routes lead into those answers: get onto the lists that match its searches, or publish the page that answers the question yourself. This blog takes the second route.

The same measurement works for any category: the questions your buyers ask, the searches the assistant runs for them, and the pages it cites. How to track whether AI recommends your business covers how to collect it, and how to measure whether AI recommendations are working covers how to read it.