Channel sheet · CH-08 · gain 5 min · logged November 18, 2026

AI MarketingDirect input

ChatGPT Panel Study Finds Purchase Intent in 6.9% of Chats

A SparkToro analysis of 41,961 U.S. panel conversations found 6.9% showed purchase intent, while 87.9% of responses in the weighted group named no brand, product or business.

By Nathan Brooks5 min read924 words

Signal notes

  1. Purchase intent appeared in 6.9% of conversations in the weighted U.S. panel analysis.
  2. The study covered 41,961 conversations from 12,321 people between Jan. 1 and Oct. 29, 2026.
  3. Among weighted responses, 87.9% named no brand, product or business.
  4. The report classified 4.9% of conversations as planned purchases with continued option research, and 2.1% as brand- or product-specific buying questions.
  5. Researchers removed 131 heaviest users, who generated 21% of all prompts.
6.9% Of ChatGPT Conversations Showed Purchase Intent In U.S. Panel
Input monitor6.9% Of ChatGPT Conversations Showed Purchase Intent In U.S. Panel — StartupStockPhotos / pixabay

By Nathan Brooks

Purchase intent appeared in 6.9%—about 7 in 100—of conversations in a U.S. opt-in panel analyzed by SparkToro. The sample covered 41,961 conversations from 12,321 people between Jan. 1 and Oct. 29, 2026.

Verso supplied the data from users who linked their ChatGPT accounts. Most reported percentages came from a statistically weighted subsample designed to match U.S. demographic characteristics.

What does the 6.9% purchase-intent figure mean?

The report divided conversations into three purposes:

  • 4.9%: The user planned to buy and continued comparing options.
  • 2.1%: The user asked about purchasing a specific brand, product or service.
  • 93.1%: The user showed no intention to purchase.

The first two categories add to 7.0% as displayed, while the report's headline and chart use 6.9%; the post does not explain the difference.

SparkToro co-founder and CEO Rand Fishkin wrote, "I suspect this is bad news for ChatGPT's advertising opportunities." He concluded that Google's AI Overviews and AI Mode, not ChatGPT, provide the main reason brands should care about AI visibility.

How often did brands appear?\n

Among weighted prompts, 86.7% did not ask for brand, product or company information. Another 6.6% asked about several brands, and 6.7% concerned one brand. Responses mentioned multiple names in 7.0% of cases and one in 5.1%, for a reported 12.1% with at least one. Separately, 87.9% of responses named no brand, product or business.

The post's takeaways called 12% of usage part of the buyer journey, although its purchase-intent chart reported 6.9%. It did not reconcile the two claims.

What did people use ChatGPT for?

The chart ranked:

  • 17.6%: Images and visual design
  • 14.4%: Information and entertainment
  • 12.7%: Writing and communication
  • 5.3%: Shopping, brands and local services

Each conversation received one category based on its primary goal. When a request produced an output, that output controlled the label: an email asking PayPal for a refund counted as writing, not shopping. Chart figures differed from the report text by less than two percentage points per category; the figures above use the chart.

Does Google search data strengthen the case?

Fishkin also cited SparkToro's December 2024 analysis of approximately 332 million Google searches. In a U.S. browser panel covering January 2023 through September 2024, he said 15% or more of searches met his definition.

Commercial intent—financial, business services or commerce—made up 14.5% of searches, though searchers might not have been ready to buy. Transactional intent accounted for 0.69%.

The Google and ChatGPT categories differ, so the source says the percentages cannot be compared directly.

How representative is this panel?

Verso offers a free tool that breaks down linked ChatGPT history. Its sign-up terms say it stores imported conversations and includes them, de-identified, in datasets licensed to research and business customers. Verso co-founder Aaron Aknin contacted the report's author and supplied the data.

The panel skewed toward younger people, women and lower-income users. Of the 12,321 members, 9,675 had demographics usable for weighting against the 2024 U.S. Census. After weighting, no usage category moved by more than 1.4 percentage points.

The researchers removed the 131 heaviest users before analysis. They averaged more than 33 prompts per active month, represented about 1% of panelists and sent 21% of all prompts. The remaining weighted group averaged 9.9 prompts from January through October.

Most panelists used mobile phones, so desktop and work-account use may be underreported. The dataset excluded API requests.

Coding used a written codebook supplied to Anthropic's Claude. Coders reviewed each conversation's title, up to six user prompts, the first few lines of ChatGPT's first few responses, and product cards or bolded names. A second coder recoded a random sample; agreement reached 90% for categories and 97% to 99% for brand tags.

The report gave no separate agreement rate for purchase intent.

How does OpenAI's September data compare?

The weighted panel placed more emphasis on images and how-to questions than OpenAI's September Signals data, which leaned toward writing and technical support. The author called the comparison fair because OpenAI probably included API requests that Verso did not track.

OpenAI's page presents a conflicting detail. Its dataset covers messages only in ChatGPT Free, Go, Plus and Pro accounts; it excludes Codex, enterprise usage and API calls, and reports global message shares. The panel covers U.S. conversations. The comparison therefore mixes scopes, even if both datasets measure message topics.

What should advertisers do now?

For brands dividing budgets between AI visibility and ads, the measured purchase-intent share is below 7%. Fishkin favors tracking Google's AI Overviews and AI Mode, and questions whether ChatGPT visibility is worth tracking unless teams also track Bing rankings and AI visibility together.

For pay-per-click teams exploring ChatGPT Ads, 6.9% is a snapshot of a consumer panel with light ChatGPT use, not a count of purchase-related conversations across ChatGPT. The panel shows what people asked better than how often they use the service.

Until panel data covers desktop, work and API activity, advertisers should pair the figure with their own ChatGPT referral and ad data before budgeting. Fishkin wants more research with Verso. SparkToro sells audience-research software; Verso's history tool is free.

For now, 6.9% is a directional measure, not a business case for moving budgets.

via searchenginejournal.com (Source)

Filed under

  • chatgpt
  • purchase-intent
  • ai-visibility
  • consumer-research
  • ai-advertising
Share this article:

More from Nathan Brooks

Nathan Brooks

Show full bio

Correspondent covering regulation, policy and their impact on marketing professionals running AI in the stack at Mart Signal.

159 articles

Bus out

‹ Previous article