Every day, millions of consumers tell AI chat what they want to buy, why, and what's holding them back. Those conversations contain some of the most detailed intent signals the industry has ever seen.

We recently analyzed 2026 year-to-date data from large language models (LLM) chat environments, search, and zero-party polling to understand how intent signals are shifting, and where brands can influence the path to purchase.

A new layer of the funnel, not a replacement

The pre-purchase digital journey starts with AI chat for over 20% of users, according to Verve’s analysis of more than a billion daily signals. Some categories were more likely to start in an LLM than others, including 37% of travel queries. Regardless of vertical, these initial prompts in AI chats are often unbranded as consumers explore category-level considerations.

A tale of two journeys

Take the auto category. A user who starts in chat faces a consideration set of 1.96 brands. This narrows to 1.86 as they use AI to curate a final shortlist, before moving to the open web.

A user who starts in search considers 4.21 brands, preferring initial research across trusted environments. After that, the user takes a curated shortlist (on average 1.43 brands) to an LLM. Here the LLM serves not as a curation layer, but as a validating sounding board for their own personal circumstances.In both scenarios, LLMs serve to curate and reinforce intent, and the competition is fierce. But despite all the important conversations around AI chat interfaces as a future retail channel or “digital shelf” for brands, many advertisers are missing an opportunity that’s already here: intent signals from LLMs.

The depth of intent

We found that users turned to AI chat throughout their journey. In sportswear, for example, 39% of prompts reflected upper-funnel informational activities, while another 37% were transactional. But the informational prompts carry buying signals that traditional search might miss. An aspiring runner might search for “Nike sneakers,” revealing in-market intent, but may also have a 10-prompt conversation about marathon running which reveals far greater depth of insight into their needs, product preferences, and purchase timings.

This doesn’t mean AI chat intent replaces search. It complements it. But now there’s an upstream layer where intent is forming, often before it ever reaches a search bar. Our data shows users typically go about six prompts deep before moving over to the open internet to convert. But this window of opportunity for brands can be short. Some convert within 48 hours, particularly for flights and electronics. Others take up to two weeks.

The share of voice rollercoaster

There’s a growing first-mover advantage for brands that can capitalize on shifting consumer patterns. Thinking back to sneakers, we saw Adidas get a 7% lift in branded search after showing up in LLM conversations. Being part of the AI conversation can clearly drive measurable downstream activity across other channels.

When brands are mentioned in LLM responses, the sentiment is overwhelmingly positive. Simply showing up is a net positive. But share of voice swings dramatically week to week. We’ve seen brands go from 12% to 62% in a matter of days. This unsettled terrain can be an advantage for brands willing to move now.

From mentions to outcomes

The measurement question has to evolve alongside the channel. Tracking the share of prompt or counting brand mentions isn’t enough. What matters is connecting those signals to business outcomes like regional sales lift, branded search impact, or downstream conversions. When someone asks an AI about a product in a specific zip code, can you tie that to real sales impact? That’s the bar.

We’re still early. We’re at about the equivalent of 1996 in the generative AI era compared to the internet revolution. The chatbot probably isn’t the end state of consumer AI. But the intent data flowing through these conversations is real, rich, and growing. Brands need to build their signal infrastructure now, rather than waiting for the ecosystem to settle.