Information seeking accounted for 20.7% of messages on person ChatGPT plans in August, nearly matching penning at 20.6%, according to OpenAI Signals data released Oct. 26. It was the archetypal duration in the series, which starts in July 2024, whenever the reported portion for seeking data was higher. Writing moved rear in forefront in September, at 21.4% versus 20.2%.
In supplement to publishing the new type of their signals (Signals V3.0), OpenAI posted a blog post concerning how users are bringing activity to ChatGPT. Their data is according to a monthly example of 1M messages from document holders alongside a declared age of 18+ on Free, Go, Plus, or Pro plans. They exclude Enterprise and Codex. In addition, they have created an automated scheme that classifies all sampled communication by topic area.
Over the months the two cover, Signals additionally shows a smaller increase in seeking data than a 2025 paper by OpenAI and Harvard researchers, which put the category at 24% of ChatGPT use by mid-2025.
Seeking Information Rose While Writing Fell
Writing cut from 34.1% of messages in July 2024 to 21.4% in September 2026. Seeking data rosy from 16.7% to 20.2% complete the identical months.
Practical guidance, which includes how-to direction and tutoring, has been the largest topic all duration since March 2025 and was 28.1% in September.
According to the release’s data dictionary, seeking data covers requests for facts, questions concerning products and services for sale, and formula or cooking help.
Work-related messages made up 32.2% of all messages in September. Among these, the portion of messages seeking data remained between 10.7% and 12.1% all duration from July 2024 through July 2026, before rising to 13.9% in August and 15.2% in September.
What The Work Activity Post Adds
The activity action article organizes work-related messages into 165 activities from O*NET, a norm example that groups tasks shared throughout jobs. It presents the data by quarter, covering Q1 2025 through Q3 2026, throughout nearly 100 countries.
Across the 74 countries alongside data in all quarter, the 10 most average activities averaged 51.1% of all country’s classified activity messages in Q3 2026, downward from 56.0% in Q1 2025. The article describes that as work-related use becoming small concentrated in the most average activities.
The post’s chart shows “Getting information,” which is one of these 165 activities. In the United States, it represented 11.6% of activity messages during the archetypal fourth of 2025. It additionally showed that it made up 9.9% of activity messages in the third fourth of 2026. These are according to a distinct classification scheme and portray lone one nation (the U.S.), so they do not measure the identical item as “seeking information” globally.
How It Compares With The 2025 Paper
Last September, I covered the OpenAI and Harvard paper on how group use ChatGPT. It reported that seeking data grew from 14% to 24% of ChatGPT use in the twelvemonth following July 2024.
Signals utilizes the exact identical classifier prompts used by the investigation study, according to the provided data dictionary, and the two categorize the identical three categories of requests under “seeking information.” Signals’ graphs additionally display the percentages of the identical period example for seeking information; 16.7% in July 2024 and 18.3% in June 2025, which is the final duration covered by the investigation paper.
The two penning sequence have small differences between them than the two seeking-information series. The percent decreased by 12 points (from 36% to 24%) in the investigation document during it decreased by 8.4 points (from 34.1% to 25.7%) in Signals during the identical period frame.
Approximately 1.1 myriad conversations were selected by the researchers using the sampling method anywhere they took lone one communication from all conversation. Each portion was afterward averaged complete 28-day periods. Signals selects 1 myriad messages per month. This publish altered former months to be accordant alongside that higher number. No logic was stated or explained inside the publish blog or the provided data dictionary concerning the difference in the seeking data percentages.
Why This Matters
In OpenAI’s own monthly data on person ChatGPT messages, about twenty percent of all messages are to appearance item up, which is approximately equal to the percent for “writing.” Product questions are included in this percentage, nevertheless since OpenAI blended all three categories (fact/recipe/product) into one category, it cannot be resolute how many were each.
Signals does track the assorted topics of conversation that happen via ChatGPT; nevertheless it did not provision any data concerning how frequently ChatGPT accessed external sources specified as websites or cited a location in its replies.
When citing percentages representing the use of ChatGPT, delight citation the origin and period frame. There was an nearly six item difference between the 2025 document and Signals concerning data seeking in mid-2025.
Looking Ahead
So far, all update in the Signals data dictionary’s changelog adds a fourth of data and revises before months. Because a fig from v3.0 may differ from the identical month’s fig in a afterward version, be certain to note the type figure whenever you contain a Signals fig in a report.
Featured Image: Tetiana Yurchenko/Shutterstock
