Only 6.9% of ChatGPT prompts involve researching a brand to buy from, according to a SparkToro study of 118,163 prompts from 12,321 US users published on 17 November 2026. Google queries met the same definition more than twice as often in SparkToro’s earlier work, so most brands should weight AI visibility budgets towards Google’s AI answers.

Key takeaways

  • SparkToro analysed 118,163 prompts in 41,961 conversations from 12,321 US ChatGPT users, sent between 1 January and 29 October 2026, and published the results on 17 November 2026.
  • 6.9% of prompts showed research into a potential brand to buy from, against more than 15% of 332 million Google queries SparkToro studied two years earlier.
  • 12.1% of ChatGPT responses contained one or more brands, products or companies.
  • Shopping, brands and local services ranked eighth of 12 use categories, at 5.7% on a per-user basis.
  • Rand Fishkin estimates Google sees purchase-journey queries at 16 times or more the rate of ChatGPT.

How did SparkToro measure ChatGPT use?

The data comes from Verso, which runs an opt-in, anonymised panel of US ChatGPT users who agreed to share their chat history. Verso cofounder Aaron Aknin offered the dataset to Rand Fishkin, who wrote up the SparkToro study. SparkToro removed 131 heavy users who averaged more than 33 prompts a month, because those 1% of panellists sent 21% of all prompts.

Claude labelled each conversation against a written codebook of 34 intents grouped into 12 categories, one label per conversation based on its main goal. A second labeller re-coded a random sample and agreed on 90% of category labels and 97 to 99% of brand tags. SparkToro then weighted 9,675 panellists with demographic records to 2024 US Census targets, which moved no category by more than 1.4 percentage points. Most figures are per user, so a handful of heavy chatters don’t set the averages.

What Americans ask ChatGPT

CategoryShare of use, per user
Images and visual design16.1%
Information and entertainment14.0%
Everyday life and admin12.2%
Writing and communication11.9%
Health and wellness10.3%
Money, work and business8.5%
Schoolwork and learning7.6%
Shopping, brands and local services5.7%
Technology support and coding4.4%
Relationships and emotional support3.4%
Chit-chat and other3.0%
Creative writing and roleplay2.9%

SparkToro defines the shopping category as deciding what to buy or where to go, such as the best laptop under $500, comparing products or finding a nearby business’s hours. Aaron Chatterji and six co-authors reached a similar picture from a different angle. Their NBER working paper, published in September 2025, found practical guidance, seeking information and writing made up nearly 80% of conversations, and non-work messages grew from 53% to more than 70% of usage.

Is ChatGPT a search engine?

Fishkin’s answer is no. “It’s not a search engine. People don’t treat it that way,” he writes, and he calls ChatGPT “a heavily-consumer app”. He compares ChatGPT’s share of buying-journey activity to Bing’s share of search, a real channel that is a fraction of Google’s. He also argues that “Google and the Google AI Overviews and AI Mode” are what make AI visibility so important.

Brands still turn up. 12.1% of responses named at least one brand, product or company, and Fishkin says ChatGPT recommended a brand a little more often than users asked for one, so some brand mentions happen in conversations the user didn’t start as shopping.

Where should AI visibility budget go?

It is tempting to treat ChatGPT prompt tracking as Google rank tracking with a new name. This study suggests the prompts those tools track (best running shoes, top CRM for startups) are a small slice of what ChatGPT users do. Paid teams saw the same gap from the other side when the first ChatGPT ads returns study found a paid ChatGPT visit earned $0.89 against $2.45 for a paid Google Search visit.

Our view at Manual Focus is that most brands should spend the larger share of their AI visibility budget on Google’s AI Overviews and AI Mode, keep ChatGPT as a smaller tracked channel, and size it the way a sensible team sizes Bing, because the purchase research happens on Google at a far higher rate and Google already shows AI Overviews on many brand searches.

Two exceptions are worth planning for. Categories that sit inside ChatGPT’s large use cases, such as health, personal finance and everyday admin, may meet buyers there in conversations that start as advice and end with a product named. And brands with physical locations or service businesses should still claim their profile through Businesses in ChatGPT, since it costs little and gives OpenAI’s own mention and click figures.

Four changes to the AI visibility plan

  1. Split the tracking budget by where buyers research. Put the larger share of prompt and citation tracking on Google AI Overviews and AI Mode, and run ChatGPT on a smaller, fixed prompt set. The channel mix simulator playbook gives a way to model the split before you commit.
  2. Rewrite the ChatGPT prompt set around real use. Swap some “best X” prompts for advice-led questions in your category’s nearest large use case, such as health, money or everyday admin, and log when ChatGPT names a brand the user didn’t ask about. The category entry points playbook helps find those situations.
  3. Report AI referrals against search, not alone. Show ChatGPT sessions and conversions next to Google organic and paid in the same report, so the board sees ChatGPT at its real size rather than as a headline.
  4. Keep one cross-engine baseline. Track which brands each engine names for the same questions, as Light Meter does across six AI engines, so a shift in ChatGPT behaviour shows up against Google rather than in isolation.

Limits of the Verso panel

The panel is US only and skews younger, female and lower-income, and most panellists used phones, so SparkToro says desktop and work-account use is likely underrepresented. API traffic isn’t captured at all, and Fishkin guesses it makes up 50 to 75% of all usage, an estimate rather than a measurement. Verso connects to OpenAI on an opt-in basis, so people who agree to share chat history may differ from those who don’t. The Google comparison uses SparkToro’s study of 332 million queries from two years ago, not a matched 2026 sample, and the split between requested and unrequested brand mentions appears only in a chart. SparkToro doesn’t report brand mention rates by category either, so a team can’t yet tell whether health or money conversations name brands more often than the 12.1% average.

Further reading

If you want help deciding where your AI visibility budget should go, our services page explains how we work.

Frequently asked questions

How many ChatGPT prompts are about buying something?

SparkToro's study of 118,163 prompts from 12,321 US ChatGPT users, published on 17 November 2026, found that 6.9% of prompts showed research into a brand to buy from. Rand Fishkin compares that with more than 15% of 332 million Google queries SparkToro studied two years earlier, and estimates Google sees purchase-journey queries at 16 times or more the rate of ChatGPT.

How often does ChatGPT mention brands in its answers?

In SparkToro's November 2026 analysis of a Verso panel of US ChatGPT users, 12.1% of ChatGPT responses contained one or more brands, products or companies. Rand Fishkin writes that responses contained a brand recommendation a little more often than users asked for one, though SparkToro gives the split only in a chart and calls the numbers fairly small.

What do people mostly use ChatGPT for in 2026?

SparkToro's per-user figures for January to October 2026 put images and visual design first at 16.1%, then information and entertainment at 14.0%, everyday life and admin at 12.2% and writing at 11.9%. Shopping, brands and local services came eighth at 5.7%. The panel is US only and skews younger, female, lower-income and mobile.

Should marketers stop tracking brand visibility in ChatGPT?

No. SparkToro's study says 12.1% of ChatGPT responses name a brand, product or company, which is a real audience. Rand Fishkin argues that Google's AI Overviews and AI Mode are what make AI visibility important, so the practical step for most marketing teams is to weight tracking and budget towards Google while keeping a smaller ChatGPT programme running.

Sources