A year ago ChatGPT was about three-quarters of gen-AI traffic. By August it was closer to half. Similarweb's tracker has been rearranging the furniture while most marketing plans stayed still. Twelve months

…see more

ago ChatGPT was roughly three-quarters of generative AI web traffic. By August it was closer to half. Gemini has taken about a quarter. Claude is near one in ten. Perplexity — the one every agency deck had a slide about — has slipped down the list. Treat the numbers as directional rather than gospel. The direction is the story. Because "AI search" was never a channel. It is four or five different systems, with different training data, different citation habits, and different willingness to name a vendor at all. Being described well inside one of them tells you very little about the other three. Which means the work a lot of teams did last year — get mentioned in ChatGPT — was a bet on a default that has since moved. What I would do instead costs nothing but an hour. Pick the three assistants your buyers actually use. Once a month, ask each one the five questions a buyer asks before they shortlist anybody. What does this company do. Who is it for. Who else should I look at. What goes wrong with it. Is it worth the money. Save the answers in one file. That file is your positioning document, written by the thing your market reads instead of your website. Where you are missing or misdescribed is a content brief. Where one assistant describes you better than your own homepage does, that is your next piece of copy. Nobody owns this channel yet. Most companies are not even reading it. When you last asked three different assistants what your company does, did the three answers agree? #AISearch #Positioning #B2BMarketing