It is easy to assume people mostly use ChatGPT to shop, or to write their emails. A large study of more than a million real conversations tells a calmer story. The single most common thing people do is seek specific information, at roughly 18 percent of queries. Next comes editing and critiquing text, then tutoring and teaching, then practical how-to advice, then personal writing. Asking about products you can buy sits far down the list, at around 2 percent.
Why the informational bias matters
Read that list again through a marketer's eyes. People are not mainly asking assistants to sell to them, they are asking to understand, to learn and to solve a problem. That is exactly the moment a brand can be useful without being pushy. When someone asks how to choose between two approaches, or how a category works, the assistant reaches for the clearest, most credible explanation it can find. If that explanation is yours, your brand rides along with the answer.
This is the practical heart of answer engine optimisation. The brands that win these moments are not the ones shouting the loudest offer, they are the ones that have published the clearest, most quotable explanation of the thing a buyer is trying to understand. We walk through the craft of it in writing content an AI can quote.
Shopping is small now, but it is the fastest-moving edge
The 2 percent who ask about purchasable products is small, but it is early. As assistants get better at browsing, comparing and even completing a purchase, that share will grow, and the brands already legible to the model will be the ones it can recommend. Being present in the informational answers today is how you earn a place in the transactional answers tomorrow.
What this means for your business
Map the real questions your buyers ask before they buy, the how and the which and the why, and answer each one plainly on a page an assistant can read. You are not writing an advert, you are writing the reference the assistant will quote. Do that consistently and you become the source it trusts in your category, which is the groundwork for every later recommendation. If you are not sure which questions you already win, the shift to assistant-led discovery explains why it is worth finding out now.