OpenAI Signals: A glimpse of the data marketers are missing?
For years, marketers have enjoyed extraordinary access to information about what consumers are searching for, how demand is changing and where opportunities exist. Google's Keyword Planner and Trends have become fundamental tools for understanding markets, identifying opportunities and informing investment decisions.
But as more consumers turn to AI platforms for information, recommendations and research, a growing proportion of that activity is happening somewhere marketers have very little visibility. The tools we've relied upon to understand customer demand are becoming less representative of the full picture.
On 26 October, OpenAI released an update to its Signals dataset, providing new insights into how people use ChatGPT, how behaviour is changing and how adoption differs across demographics and geographies.
It's not yet the equivalent of Google Keyword Planner, but could this be an early indication that a world with more information for marketers is on the way?
How is the usage of ChatGPT evolving?
The dataset covers ChatGPT usage between July 2024 and September 2026, and whilst the dimensions presented are still somewhat vague, several findings stand out.
1. Users are using AI increasingly with commercial intent
Between July 2024 and September 2026, "seeking information" has increased by c.4ppts to 20%. To caveat that we do not have the exact prompts to know what is being searched, but this is likely the pool with which messages looking for information about products and services or commercial intent as we define it would sit.
Note that “practical guidance” has grown to be the largest share at 28% of messages.
For marketers, the significance is that ChatGPT is increasingly a place people go to ask questions and seek guidance, (conduct upper funnel research) not simply a tool for completing tasks.
2. The audience is broadening
Among users who have disclosed their age, the proportion of messages from 18–24-year-olds has fallen from 39% to 31%, while the share from users aged 65 and over has tripled, albeit from a much smaller base.
Meanwhile, messages associated with traditionally feminine first names have increased from 42% to 52% of messages where name-based classification is possible. These are shares of messages rather than user population figures, but they suggest that ChatGPT usage is broadening beyond its original audience.
3. How people use AI varies considerably
OpenAI distinguishes between people asking for information, doing something and expressing themselves. The balance differs between work and personal use, with work-related interactions more focused on completing tasks, while personal usage is more heavily weighted towards asking questions.
This is an important distinction for marketers as they seek to understand how AI fits into the customer journey. Not every prompt is a search, and not every question represents commercial intent.
Source: OpenAI Signals v3.0, October 2026.
Is Google search still reliable as the primary indicator of market demand?
For many years, Google has given marketers an extraordinary view of consumer demand. We could assess how frequently consumers searched for particular products, services or categories, understand seasonality and identify emerging opportunities. This information has been fundamental to how businesses plan their marketing activity and allocate investment.
We are now entering a period where some of that visibility is disappearing. Platforms such as Profound and Peec AI are helping businesses understand how frequently their brands appear in AI-generated answers, which competitors are being recommended and which sources influence those responses. Some also offer estimated prompt demand, but in reality this is a calculated field and not actually a true representation of what is going on in the market. It is definitely not yet close to the breadth of demand data available through traditional search.
Put simply: Google keyword trends are no longer a good enough indicator of market demand.
The shift towards AI discovery is becoming visible in the data
We're now beginning to see evidence of what many have suspected would happen, and what we explored in our earlier article on the future of ecommerce and agentic AI: product discovery and research are beginning to shift away from traditional search engines towards LLMs.
At Coppett Hill, our analysis of search demand across different categories is beginning to reveal some interesting patterns. In particular, we're seeing declines in Google search volumes across certain upper-funnel query types, where consumers have traditionally started their research and discovery journeys.
Of course, declining search volumes alone cannot prove that those questions have moved to AI platforms. However, alongside the rapid growth in conversational AI usage, we're beginning to see patterns consistent with that shift becoming large enough to register in the data.
Our analysis indicates declining Google search demand across both B2C and B2B client segments
Moreover, we're specifically seeing impact on product discovery and inspiration (or "upper funnel") searches.
This poses an important challenge for marketers. A reduction in Google search volumes doesn't necessarily mean a reduction in underlying consumer demand; it may simply mean that consumers are researching products and services somewhere else. Without understanding how different types of searches are changing, businesses risk misinterpreting market demand and making investment decisions based on an increasingly incomplete picture.
A question to the reader 💬
How well can you distinguish between a genuine reduction in demand and a change in where that demand is being expressed?
It's important, however, not to confuse the growing significance of AI with the declining relevance of traditional search. Google remains an enormously important marketing channel, and consumers will continue to use it for discovery, comparison and purchasing.
In fact, if I had to make a bet, I'd suggest that Google's own AI, Gemini, will be one of the long-term winners in this transition. But that's a topic for another article.
How can marketers retain visibility of demand and draw actionable insights?
OpenAI has made no commitment to provide any further level of commercial detail. The current Signals dataset is aggregated and doesn't provide individual prompt volumes, brand-level insights or the ability to measure demand for specific products. Nevertheless, the release is a positive step towards greater transparency, and potentially a sign of more useful intelligence becoming available to marketers over time.
For marketers, however, the challenge is more immediate. As customer behaviour evolves, businesses cannot wait for better data to become available. They need to:
Understand how customer acquisition is changing. Develop a clearer understanding of how the roles of marketing channels are evolving, where they should be investing
Build a more connected view of marketing performance. Look beyond measuring individual channels. Instead build a holistic view by connecting the available data across traditional search, AI visibility, website behaviour and ultimately commercial outcomes.
At Coppett Hill, we help businesses piece together that picture. By connecting data across marketing, sales and customer outcomes, we help management teams understand what's changing, where the opportunities are and how to make better commercial decisions.
If you want to understand how these changes are affecting your own website, customer acquisition and wider business performance, get in touch here. We can help you identify where search behaviour is changing, understand how your visibility across Google and AI platforms is evolving, and connect that insight to commercial performance so you can make informed decisions about where to invest next.
The views expressed in this article are those of the author and are intended to inform and provoke discussion. They should not be treated as professional advice or relied upon as a substitute for advice specific to your circumstances.
