How to Track Whether ChatGPT Recommends Your Brand

To track whether ChatGPT recommends your brand, ask it the same set of shopper questions many times, without naming your brand, and measure how often you appear. A single answer tells you almost nothing, because AI tools rarely give the same list of brands twice.

Track two numbers against your competitors: how often you are mentioned at all, and how often you are the top pick. Run the same questions every month and read the trend, not your position in any one answer.

Why can’t you just ask ChatGPT once?

Because the answer changes almost every time. In a study by SparkToro and Gumshoe, 600 volunteers ran 12 prompts through ChatGPT, Claude and Google’s AI search tools a combined 2,961 times. There was less than a 1 in 100 chance that ChatGPT or Google’s AI would give the same list of brands in any two responses, and the order of a list repeated closer to 1 in 1,000 runs (SparkToro).

So a screenshot of one answer, good or bad, is an anecdote. The same study concluded that visibility percentages measured across many prompts are a reasonable proxy for how brands actually show up in AI answers. SparkToro discloses that its co-author works at Gumshoe, which sells AI tracking.

Why does AI visibility matter for sales?

Because shoppers already ask AI what to buy, and the traffic it sends is growing fast. OpenAI says hundreds of millions of people use ChatGPT to find, understand and compare products, and in November 2025 it added shopping research, which builds a buyer’s guide from a description of what someone needs (OpenAI).

Adobe measured traffic to US retail sites from generative AI tools up 693.4% year on year over the 2025 holiday season, November 1 to December 31, and up 670% on Cyber Monday (Adobe). A brand that is missing from those answers is missing from the shortlist before a shopper ever reaches a search page.

Which AI visibility metrics should you track?

Five, in this order of importance. The first two are the ones to report; the rest explain them.

Metric	What it answers	How to calculate it	Watch out for
Mention rate	Are we in the consideration set at all?	Answers that name you ÷ all answers	Counts a passing mention the same as a recommendation
Top-pick rate	How often are we the recommendation?	Answers where you are the first or named-best choice ÷ all answers	Needs a consistent rule for what counts as the top pick
AI share of voice	How do we compare with competitors?	Your mentions ÷ mentions of every brand in the answers	Shifts when the AI names more or fewer brands per answer
Average position when named	When we appear, how high?	Mean list position across answers that name you	Order is the least stable part of an AI answer; read it last
Cited sources	What is the AI reading about our category?	Domains and channels cited in the answers, and whether they name you	Some engines cite sources more often than others

Track every metric per engine as well as combined. ChatGPT and Google’s AI draw on different sources: in OtterlyAI’s analysis of more than 100 million AI citations, ChatGPT produced 4.4% of the YouTube citations it observed, against 36.6% for Google’s AI Overviews (OtterlyAI). A brand can lead on one engine and be missing from the other.

How do you build a fair set of questions?

Use the questions a shopper asks before they have a brand in mind, and never name your own brand in them. A prompt that names you measures whether the AI recognizes you, not whether it recommends you.

Cover the ways people actually shop the category. An illustrative set for a running shoe brand:

Question type	Example	What it tests
Best in category	“What are the best running shoes for beginners?”	Whether you make the default shortlist
Use case	“Which running shoes are good for flat feet?”	Whether you own a specific need
Price	“What are good running shoes under $100?”	Whether you show up at your price point
Comparison	“Is it worth paying more for carbon-plated shoes?”	Whether the AI uses you as an example
Problem	“Why do my shins hurt after running, and what shoes help?”	Whether you appear when the shopper starts from a problem

Group the questions by topic, so you can see where you win and where a competitor does. Then keep the set fixed. If you change the questions, you can no longer compare this month with last month.

How many times should you ask each question?

More than once per engine, every time you measure. Because answers vary from run to run, each question is a sample, not a verdict. The number that means something is the percentage across the whole set: of all the times shoppers’ questions were asked, how often were you named, and how often were you the pick.

Run every question on every engine you care about, on the same schedule, with the same settings. Read changes of a few points with care. A move that persists for two or three months is a trend; a move in one month may be noise.

How often should you measure AI visibility?

Monthly. The models, the sources they cite and the content creators publish all change faster than a quarterly report can follow, and a weekly report mostly measures noise. A monthly run on a fixed question set gives you a trend line that moves when something real changes, such as a competitor’s creator push or a model update.

Compare yourself with the same competitors each month. The brands the AI names alongside you are your real AI competitive set, and they are not always the rivals you would list yourself.

What do you do when a competitor is the top pick?

Read the sources. The cited domains and channels show what the AI is learning from in that topic. If the answers lean on a handful of YouTube channels or reviews that mention a competitor and not you, those are the places to earn a mention.

Creator content is often the fastest way in. In Ahrefs’ study of 75,000 brands, YouTube mentions were the signal most strongly correlated with a brand appearing in AI answers (Ahrefs). Answer engine optimization runs on creator content covers how to brief creators so their videos answer the questions your buyers ask, and competitor influencer analysis shows how to find the creators already driving a rival’s attention. Then rerun the same questions next month to see whether the gap closed.

AI visibility is one view of the same race that share of voice measures on social. A brand that is gaining on TikTok and Instagram but missing from AI answers is winning attention it is not yet converting into recommendations.

Can you track AI visibility without building it yourself?

Yes. On SocialNative.ai, AI Visibility asks ChatGPT and Gemini the questions shoppers ask about your category every month, without naming your brand, then shows how often you are named, how often you are the top pick and which sources the answers cite, next to every competitor the AI names.

See how AI ranks your brand