For years, ChatGPT was the office drafting machine: emails, reports, formulas. OpenAI’s new Signals data shows that version of the product is fading fast.
Writing falls, practical guidance takes the top spot
OpenAI Signals v3.0, released on October 26, classifies one million adult consumer messages each month and breaks them into seven topics. Writing fell to 21.4% of requests in September 2026, down from 34.1% in July 2024. Work-related use dropped from 51.6% to 32.2% over the same period.
Practical guidance has been the largest topic every month since March 2025, rising from 26.1% to 28.1%. Seeking information climbed from 16.7% to 20.2%, while multimedia jumped from 2.0% to 8.8%. Technical help collapsed from 13.5% to 4.6%, likely as coding moved to dedicated agents and developer tools.
Why ad buyers should pay attention now
ChatGPT sells advertising only on its Free and Go plans, and those are exactly the plans that have moved furthest from work. In September 2026, about three-quarters of Free requests and 57.6% of Go requests were not work-related. As PPC Land notes, the release itself “says nothing about advertising,” but the ad audience is being reshaped underneath it.
- Writing: 34.1% to 21.4%
- Practical guidance: 26.1% to 28.1%
- Multimedia: 2.0% to 8.8%
- Self-expression: 4.0% to 8.9%
- Technical help: 13.5% to 4.6%
- Work-related requests: 51.6% to 32.2%
The intent shift is even starker. Across all requests, expressing rose from 11.7% to 29.1%, while asking fell from 45.7% to 36.9% and doing fell from 42.7% to 33.9%. Users are increasingly talking to ChatGPT, not just tasking it.
What this means for your next campaign
For performance marketers, this is not a curiosity. The audience is becoming older, more evenly split by gender and more interested in advice and images. MediaPost reported that OpenAI added a visual ad format and measurement tools, letting advertisers “tell their story through images” while keeping ads separate from generated pictures.
The practical play: stop planning ChatGPT campaigns around drafting and debugging. Build for the advisory and visual moments that now dominate consumer use. At the same time, sensitive-conversation filters can limit delivery, so underdelivery is a real trade-off and measurement partners matter.
Source: PPC Land
