OpenAI released version 3.0 of its OpenAI Signals data on 26 October 2026, and its files show work messages about the values or prices of goods and services rising from 0.26% to 0.69% of US work-related ChatGPT messages between the first quarter of 2025 and the third quarter of 2026. By our arithmetic, that share grew about 2.6 times.
On 27 October 2026, the SEO analyst Kevin Indig posted on X a breakdown of the update, saying he “went through every file and found a few surprises.” We downloaded OpenAI’s CSV files and checked the figures used here against them.
Business Pill · AI agents explained
An evergreen Business Engineer explainer on AI agents. It explains the general idea and does not discuss OpenAI’s Signals data.
The key insight: As we read it, OpenAI’s files show people bringing more questions about goods, services and prices to ChatGPT at work, and fewer requests to research a subject or write, as shares of individual-plan use between early 2025 and late 2026.
OpenAI says the release covers “165 work activities across nearly 100 countries, revealing how consumers use ChatGPT for work.” It maps ChatGPT messages to O*NET’s Intermediate Work Activities and is reported quarterly from the first quarter of 2025 through the third quarter of 2026.
The data has limits that OpenAI states itself. It reflects only messages from ChatGPT Free, Go, Plus and Pro accounts, and it “does not include enterprise messages or Codex use.” OpenAI says the work data “likely underrepresents business and technical use cases.”
OpenAI also reports that work use spread out: across 74 countries observed every quarter, the ten most common activities fell from 56.0% of each country’s classified work messages in Q1 2025 to an average of 51.1% in Q3 2026.
Share of US work-related ChatGPT messages by O*NET work activity, Q1 2025 vs Q3 2026, from OpenAI’s Signals v3.0 data files (released 26 October 2026). The data covers individual ChatGPT plans only, excludes enterprise and Codex, and reports shares, not message counts.
What Was Posted on X
OpenAI just updated its dataset on how people use ChatGPT.
I went through every file and found a few surprises:
1. Classic Google query types are now about half of all messages
– 48% of all messages were Google-style questions in September 2026, up from 43% in July 2024.
– The 48% breaks down into 20% simple searches (facts, products, recipes) and 28% Practical Guidance (how-to, tutoring, health, creative ideas).
– OpenAI sorts every message into work or personal. Google-style questions make up 53% of personal messages and 38% of work messages (September 2026).
– Personal messages grew from 48% to 68% of all messages since July 2024.
– Most of the rise comes from personal messages taking a bigger share, because they’re search-heavy.
2. Commercial research at work grew 75%
Among US work messages (Q1 2025 vs Q3 2026):
– “Obtain information about goods or services”: 1.7% โ 2.9%
– “Determine values or prices of goods or services”: 0.3% โ 0.7%
– “Gather information from physical or electronic sources”: 1.2% โ 1.8%
Surprised me: researching a subject for work fell by up to two thirds (technology -66%, history and social issues -65%, health -42%). Product and price research grew.
3. Countries that use ChatGPT more use it more for search
– The quarter of countries with the highest ChatGPT use per capita put 24% of messages into simple searches in Q3 2026, vs 16% for the lowest quarter.
– Across ~150 countries, the correlation grew from 0.30 to 0.74 (Q1 2025 vs. Q3 2026).
4. Personal use grows faster than work
– OpenAI’s classifier tags every sampled message as work-related or not (consumer plans only).
– Work went from 52% of messages in July 2024 to 32% in September 2026. The data only shows shares, so work messages could still be growing in absolute numbers.
In September 2026, the share of messages that were work-related was:
– Free: 25%
– Go: 42%
– Plus: 51%
– Pro: 68%
Source: OpenAI Signals v3.0, announced in “ChatGPT’s expanding world of work”
159 countries, 27 million sampled messages, July 2024 to September 2026
Kevin Indig posted this on X on 27 October 2026, saying OpenAI had updated its dataset on how people use ChatGPT and that he went through every file. He reports that 48% of messages were Google-style questions in September 2026, that commercial research in US work messages grew 75%, and that countries with higher per-capita use put more messages into simple searches. The piece checks his share figures against OpenAI’s published files and attributes his country analysis and labels to him.
Product and Price Questions at Work
Share of US work-related ChatGPT messages that OpenAI classifies as “Determine values or prices of goods or services,” Q1 2025 and Q3 2026, from OpenAI’s Signals v3.0 data files. The 2.6x ratio is our arithmetic; the data covers individual plans only and reports shares, not counts.
In OpenAI’s US file, “Obtain information about goods or services” rose from 1.69% to 2.92% of work-related messages between Q1 2025 and Q3 2026. “Determine values or prices of goods or services” rose from 0.26% to 0.69%, and “Gather information from physical or electronic sources” from 1.16% to 1.83%.
Together, by our arithmetic, those three activities went from 3.12% to 5.44% of US work messages, a rise of about 75%. Indig reached the same 75% figure and calls it commercial research at work.
Several research activities moved the other way. In the same US file, “Research technology designs or applications” fell from 1.51% to 0.51%, “Research historical or social issues” from 1.43% to 0.51%, and “Research healthcare issues” from 1.54% to 0.89%. By our arithmetic those are falls of about 66%, 65% and 42%.
Personal Use Now Dominates
Three shifts in OpenAI’s Signals v3.0 data: price questions’ share of US work messages (about 2.6x, our arithmetic), messages not related to work (68% in September 2026) and writing (21% in September 2026).
OpenAI’s monthly file shows messages not related to work rising from 48.4% of all sampled messages in July 2024 to 67.8% in September 2026. Work-related messages fell from 51.6% to 32.2%.
Indig notes that “the data only shows shares, so work messages could still be growing in absolute numbers.” The share of work messages also differs by plan: in September 2026, OpenAI’s file puts it at 24.7% on Free, 42.4% on Go, 50.8% on Plus and 68.1% on Pro.
Questions Versus Writing
Two of OpenAI’s topic categories, “Seeking information” and “Practical Guidance,” made up 20.2% and 28.1% of messages in September 2026, up from 16.7% and 26.1% in July 2024. By our arithmetic that is 48.3% combined, up from 42.8%; Indig groups the two as “Google-style questions.”
Writing fell from 34.1% to 21.4% of messages over the same period, and Technical help from 13.5% to 4.6%, a category the files measure without Codex or enterprise use. In September 2026, the two question categories made up 53.0% of non-work messages and 38.1% of work messages, by our arithmetic.
Indig’s Country Finding
Indig also looked at countries. By Kevin Indig’s analysis, the quarter of countries with the highest ChatGPT use per capita put 24% of messages into simple searches in Q3 2026, against 16% for the lowest quarter, and the correlation across about 150 countries rose from 0.30 to 0.74 between Q1 2025 and Q3 2026. We did not recompute these country figures.
As we read it, the clearest US work shift in the files is toward finding out about goods and services and what they cost: by our arithmetic the three activities went from 3.12% to 5.44% of work messages.
As we read it, the files also record ChatGPT becoming mostly a personal tool: OpenAI’s data puts messages not related to work at 67.8% in September 2026, up from 48.4% in July 2024.
As we read it, the limits matter as much as the trend: OpenAI says the data excludes enterprise and Codex and likely underrepresents business and technical use, and it reports shares, not counts.
“This data gives researchers a new way to study AI’s role in the economy, with new information on the work activities people bring to ChatGPT.”
Three Implications
FOR SEARCH AND SHOPPING BUSINESSES OpenAI’s files show the share of US work messages about goods, services and prices rising between Q1 2025 and Q3 2026; they do not show purchases or which products were asked about.
FOR RESEARCHERS OpenAI says the data maps ChatGPT messages to 165 O*NET work activities across nearly 100 countries, quarterly from Q1 2025, and invites researchers to connect it with other sources.
FOR READERS OF THE NUMBERS Indig’s country correlation and his labels are his own analysis; the shares in this piece come from OpenAI’s published files, with our arithmetic labelled.
The framework puts it this way: “They are not the same curve, and the gap between them — what the platform makes possible versus what the product surfaces — has a name.”
As we read it, OpenAI’s files are a record of the user side of that gap: the share of messages asking for information and guidance rose while writing and technical help fell as shares. The framework was published in June 2026 and does not discuss OpenAI’s Signals data.
What Is Not Established
OpenAI’s files report shares of sampled messages, not counts, so they do not show whether the number of work messages or of price questions rose or fell. They exclude enterprise plans and Codex, which OpenAI says likely underrepresents business and technical use.
The files do not say which products or prices people asked about, or whether those conversations led to purchases. “Google-style questions” and “commercial research” are Indig’s labels, not OpenAI’s categories.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
OpenAI’s Signals v3.0 files show US work messages about goods, services and prices taking a larger share of ChatGPT work use between Q1 2025 and Q3 2026, while subject research and writing shrank as shares, in data that covers individual plans only.
A note on sourcing. Figures are from OpenAI’s Signals v3.0 release (26 October 2026): its release page, its data page and its published CSV files, which we downloaded and read on 27 October 2026. Activity names are O*NET Intermediate Work Activity titles. Figures labelled “by our arithmetic” are ours; the country findings are Kevin Indig’s analysis, posted on X on 27 October 2026, and were not recomputed. This is not investment advice.
Short explainer · Anthropic and the GSA
A vertical news Short narrated from the published text of this piece and related FourWeekMBA pieces.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
