EMARKETER’s Sky Canaves: AI is Reshaping the Middle of the Funnel
Despite expectations of shoppers turning more to artificial intelligence agents, consumers aren’t discovering new brands through AI. And they aren’t completing purchases through these AI channels either. What they are doing is using it to research, compare, and validate. And that mid-funnel behavior carries significant implications for how brands need to show up.
“Consumers have told us that they don’t really see AI as a channel for discovery and awareness of new products, and they’re not really buying through AI. But what they do is seek out AI product recommendations, compare and contrast products, and look for the best deals,” Sky Canaves, principal retail analyst at EMARKETER, told Beet.TV contributor David Kaplan at EMARKETER’s Future of Digital event in New York. “When they do that, they don’t just take AI answers at face value. They tend to go and conduct their own additional research to validate.”
That validation behavior means AI is adding touchpoints to the consumer journey, not replacing them. That’s creating both opportunity and complexity for brands trying to influence purchase decisions.
Three imperatives for influencing agents
Canaves outlined the strategic framework brands need to influence AI recommendations across three layers.
First, invest in the content that LLMs and humans both trust. That means turning to authoritative sources, user reviews, and human-created material that signals credibility. Second, ensure product data is fully readable by AI systems, including extensive product attributes, use case mapping, FAQs, and clear image and video descriptions.
“If any of this information is inaccurate, inconsistent, or incomplete, there’s a real risk that a brand or product won’t show up in AI conversations at all,” Canaves said.
The third layer is paid AI advertising. It’s still nascent, but worth experimental budget allocation to begin understanding what works before the category matures.
Autonomous commerce is overhyped
The vision of agents independently completing purchases on consumers’ behalf remains a future state that consumer behavior doesn’t yet support.
“Consumers aren’t really willing to give up their decision making authority when it comes to spending money,” Canaves said. “The real opportunities are really in that mid-funnel content, providing richer and more extensive product data that satisfies LLM needs and ensuring that brand messaging is consistent across channels wherever a consumer may encounter a brand.”
The exception Canaves sees emerging involves retailers with deep consumer trust and purchase history, such as grocery replenishment and low-stakes recurring purchases where autonomous buying makes intuitive sense.
More touchpoints, harder measurement
When consumers use AI in their research process, they engage with more brand touchpoints than when they don’t. It’s a dynamic that creates genuine opportunity alongside a measurement challenge.
“When consumers use AI, they’re actually engaging with more touch points than when they don’t use AI. That gives brands more opportunities to influence purchasing decisions, but it also demands more content and makes measurement and attribution a bit more of a challenge,” Canaves said.
The path to purchase over the next few years will be increasingly AI-mediated and AI-assisted, but human decision-making authority stays intact.
“Brands will still need to provide the types of content and experiences and products that human consumers want and demand,” Canaves said.
You’re watching Beet.TV’s coverage of EMARKETER’s 2026 The Future of Digital summit. For more videos from this summit, please visit this page.
Despite expectations of shoppers turning more to artificial intelligence agents, consumers aren’t discovering new brands through AI. And they aren’t completing purchases through these AI channels either. What they are doing is using it to research, compare, and validate. And that mid-funnel behavior carries significant implications for how brands need to show up.
“Consumers have told us that they don’t really see AI as a channel for discovery and awareness of new products, and they’re not really buying through AI. But what they do is seek out AI product recommendations, compare and contrast products, and look for the best deals,” Sky Canaves, principal retail analyst at EMARKETER, told Beet.TV contributor David Kaplan at EMARKETER’s Future of Digital event in New York. “When they do that, they don’t just take AI answers at face value. They tend to go and conduct their own additional research to validate.”
That validation behavior means AI is adding touchpoints to the consumer journey, not replacing them. That’s creating both opportunity and complexity for brands trying to influence purchase decisions.
Three imperatives for influencing agents
Canaves outlined the strategic framework brands need to influence AI recommendations across three layers.
First, invest in the content that LLMs and humans both trust. That means turning to authoritative sources, user reviews, and human-created material that signals credibility. Second, ensure product data is fully readable by AI systems, including extensive product attributes, use case mapping, FAQs, and clear image and video descriptions.
“If any of this information is inaccurate, inconsistent, or incomplete, there’s a real risk that a brand or product won’t show up in AI conversations at all,” Canaves said.
The third layer is paid AI advertising. It’s still nascent, but worth experimental budget allocation to begin understanding what works before the category matures.
Autonomous commerce is overhyped
The vision of agents independently completing purchases on consumers’ behalf remains a future state that consumer behavior doesn’t yet support.
“Consumers aren’t really willing to give up their decision making authority when it comes to spending money,” Canaves said. “The real opportunities are really in that mid-funnel content, providing richer and more extensive product data that satisfies LLM needs and ensuring that brand messaging is consistent across channels wherever a consumer may encounter a brand.”
The exception Canaves sees emerging involves retailers with deep consumer trust and purchase history, such as grocery replenishment and low-stakes recurring purchases where autonomous buying makes intuitive sense.
More touchpoints, harder measurement
When consumers use AI in their research process, they engage with more brand touchpoints than when they don’t. It’s a dynamic that creates genuine opportunity alongside a measurement challenge.
“When consumers use AI, they’re actually engaging with more touch points than when they don’t use AI. That gives brands more opportunities to influence purchasing decisions, but it also demands more content and makes measurement and attribution a bit more of a challenge,” Canaves said.
The path to purchase over the next few years will be increasingly AI-mediated and AI-assisted, but human decision-making authority stays intact.
“Brands will still need to provide the types of content and experiences and products that human consumers want and demand,” Canaves said.
You’re watching Beet.TV’s coverage of EMARKETER’s 2026 The Future of Digital summit. For more videos from this summit, please visit this page.
Series Videos
