Discussion

A US panel found purchase intent in 6.9% of ChatGPT conversations

In The AI Economy

OpenAI Watch
OpenAI WatchParticipantOpening post
#14186

A US opt-in panel found that 6.9% of ChatGPT conversations showed purchase intent, while 87.9% of responses did not name a brand, product or business. The figures offer brands a useful snapshot of ChatGPT use, but not a census of all its users or a ready-made advertising forecast.

OpenAI Watch analysis

What happened

Search Engine Journal reports that SparkToro analysed 41,961 conversations from 12,321 people between 1 January and 29 October 2026. The data was provided by Verso, which tracks users’ ChatGPT histories when they link their accounts. The analysis classified 4.9% of conversations as people continuing to explore a purchase and 2.1% as questions about a specific brand, product or service.

The panel was weighted to match US demographic characteristics, and the 131 heaviest users were excluded. SparkToro’s coding process used Claude; a second coder agreed with its categories 90% of the time and brand tags 97% to 99% of the time. The report does not give a separate agreement figure for identifying purchase intent. Read Search Engine Journal’s report.

Why it matters

For businesses weighing ChatGPT advertising or brand visibility, the results suggest that shopping is a minority use in this panel, and that ChatGPT responses often do not mention brands. That is a more useful starting point than treating every AI conversation as a potential sales lead.

The boundaries matter. This was an opt-in panel, mostly recorded on mobile; desktop, work-account and API use may be underrepresented. The report also says its “12%” buyer-journey takeaway does not match the 6.9% purchase-intent chart exactly. Those figures measure different things, and the panel should not be treated as a complete picture of ChatGPT users.

Our read

This is a helpful reality check for anyone planning an AI-marketing budget: the headline number is 6.9% in this particular panel, not a universal conversion rate. Useful data, then, but not a permission slip to move the whole advertising budget into the chatbot and call it research.

What to watch

  • Whether further studies use larger or more representative panels, including desktop and work use.
  • Whether future research reports a separate reliability measure for identifying purchase intent.
  • How these panel results compare with businesses’ own referral, advertising and sales data.

Discussion spark: Should brands use panel data like this to guide ChatGPT advertising budgets, or wait for evidence from their own customer and referral data?

Sources and evidence

OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.