Buying something used to be simple. Now a single purchase touches half a dozen surfaces before it is done. A shopper opens an AI chatbot to figure out what they want, jumps to Google to search the specifics, clicks through to a retailer, doubles back to ChatGPT to compare two options, then finally hits buy. Often across different sessions, different devices, different days.
One line captures it: Google is where people act, and AI is where people think.
That is the headline finding from Similarweb, which tracked US search demand across both channels from January to August 2026. The same split surfaced in every category it examined.
The act-versus-think divide
Arctic travel shows it most cleanly. On Google, people search bookable specifics: "Glacier Bay cruise", "Arctic igloo resort", "Arctic sightseeing flights". Those searches rose 46%. They already know the category and are performing a concrete action within it.
On AI, the same travellers are still deciding whether and how to go at all. What is the best seven-day Alaska itinerary? Cruise or a land-based trip? Those queries grew 8.8%. Google rewards operators known well enough to be searched by name. AI is where the undecided, comparative questions live.
Apparel makes the same point sharper. "Will this actually fit me" queries jumped 70% on AI against just 5% on Google. The reason is obvious once you see it. Google can point you to a brand's size chart. It cannot tell you whether the jumper will suit your body. That question is moving to AI, and fast.
Same pattern, three categories
The divide held across every vertical.
In adventure tourism, AI demand rose 66.8% while Google slipped 6.1%. AI growth was led by African travel planning and detailed itineraries, the kind of open-ended research that plays to a chatbot's strengths.
Apparel and accessories climbed 82.5% on AI and 21.3% on Google. The two channels even move to different rhythms. Google follows the old seasonal wardrobe refresh, spiking in March then again through early summer. AI reflects a slower, more research-heavy back-to-school journey that now peaks in August.
Sporting goods was the standout. AI demand soared 102.2%, Google grew 42.5%. AI builds early and peaks in June as summer activity kicks in, while Google holds flat through spring and spikes in August for back-to-school and fall gear. Footwear advice split hardest of all: Google brand searches fell 11% while AI fit and comfort questions rose 17%.
The thread runs through all three. AI grows faster, and it captures the earlier, messier part of the decision, before anyone is ready to type a brand name into a search box.
This is not AI killing search
The obvious conclusion, that AI is eating Google's lunch, does not survive contact with the numbers. Traffic to AI platforms grew 70% year on year. But social traffic rose 7% over the same period, and traditional search still grew 2%. Nobody is being replaced.
The pie is getting bigger and more fragmented at once. That is the part that should concentrate minds. Watching a single channel now tells you far less than it used to, because intent moves between surfaces before it ever converts. A dip in Google volume for a category might not mean falling demand. It might mean the thinking has migrated to AI, with the buying still to follow.
Win both ends, or half the funnel
The takeaway for brands is uncomfortable but clear. You have to win both ends of the same journey.
Show up in Google when someone already knows what they want and is ready to act. Show up in AI's answers while they are still working out whether they want it at all. Increasingly, that second job comes down to having structured, accurate data an AI can actually reference, so you become the name it suggests long before a shopper types a brand into Google.
Miss either end, and you are only capturing half the funnel.
