AI search shift: is it killing demand, or just moving it?

The AI search shift has marketers bracing for the worst. AI overviews, zero-click results, and chatbots that answer the question before anyone reaches your website are feeding one fear: that AI search will kill web traffic and hollow out demand.

Our data tells a more interesting story. After analysing more than 41 million inbound calls over 12 months, we found that traffic may be thinning, but intent is thickening. Demand isn't disappearing, but it is moving, and a growing share of it is landing on the phone.

This post draws on our AI search shift whitepaper and the live webinar discussion between Daniel, our Chief Customer Officer, and Shamsul Chowdhury, SVP Paid Media at Zeno Group. It covers what's changing, what's breaking and what to do about it.

Here's the sneak peek

Whitepaper

What 41 million phone calls tell us about the AI search shift

Get the whitepaper

Free webinar

Unpacking the impact of the AI search shift on buyer behaviour

Watch on-demand

Get the whitepaper

Back

Watch on-demand

Back

What is the AI search shift?

The AI search shift is the move from buyers researching on a search results page and clicking through to your website, to buyers researching inside AI tools such as ChatGPT, Perplexity and Gemini, and in AI overviews, before they ever contact a brand. Discovery and comparison now happen earlier, often somewhere you can't see. The whitepaper's central thesis is simple: AI handles discovery, and the phone handles commitment.

Is AI search killing demand? Traffic is thinning, but demand is concentrating

Fear sells headlines. Shamsul compared the current panic to the moment Google said cookies were going away: much of the fear comes from advertisers being unready for change. Web traffic is falling in places, he said, but that reflects how people now discover brands – across social, video, maps, AI agents and phone calls – not a collapse in interest.

The call data backs that up. Between January 2025 and June 2026, the visit-to-call rate rose 31%, from 1.59% to 2.04%. Call volume grew 24% in the first half of 2026 compared with the first half of 2025. Fewer people may be visiting, but the ones who do are more likely to pick up the phone.

As Daniel put it, "while traffic might be thinning, demand is not disappearing."

Why phone calls are the clearest intent signal

Clicks tell you someone arrived. Calls tell you someone is serious. Shamsul's take was that picking up the phone means giving your time, which is a finite resource, and that is "a huge indicator of a quality interaction" compared with someone scanning a page and bouncing.

The numbers fit. More than 70% of calls from paid search, organic search and AI referrals last longer than a minute. Yet Daniel noted that many marketing teams still don't do enough with what happens on the call, leaving it to sales teams who own it. That's a big missed opportunity for anyone responsible for performance.

Paid search and AI referrals: where high-intent calls come from

Paid search isn't the casualty many expected. Across 11.8 million PPC-attributed calls, paid search generated calls at 3.08% of visits, against 1.76% for organic search – around 1.75 times the rate. Shamsul went further, arguing that paid search is a beneficiary of AI as LLM platforms introduce ads. He also noted that AI overviews appear more often on informational queries than transactional ones, so query intent decides how exposed you are.

AI referrals are the fastest-growing source. They grew 1,989% between January 2024 and April 2025, with ChatGPT accounting for 78.5% of that traffic. The base was small, at 68,247 landings a month in April 2025, but the quality is worth watching. Across 16,381 AI-referred calls, 73.6% lasted longer than a minute, and events occurred at a rate of 1.45%, compared with 0.99% for PPC.

What breaks when buyers arrive already informed

Buyers now arrive having researched, compared and priced up options in an LLM. The conversation shifts from "what does this do?" to "why should I pay for it?"

The whitepaper shows it clearly. In Q2 2026, price and budget mismatch accounted for 31.6% of non-converting sales leads, up 77% year on year. Product understanding fell by 55%, to around 8%. Across 12 months, 260 of every 350 verified sales-intent calls didn't convert.

Daniel's explanation: callers increasingly phone to be closed, but scripts and sales teams are still in education mode. The consumer journey is optimised in sections by different people, and nobody owns the handoff from marketing investment to closed revenue. Shamsul added a warning – a mildly educated buyer can be harder to convert than one with no prior knowledge, because they may arrive with preconceptions that are out of date or wrong.

Why the click is a weaker proxy for intent, and who owns the fix

AI platforms are largely a black box for attribution. Shamsul suggested using third-party tools directionally, to see whether you're showing up more and moving the right way. No attribution model is complete today, and waiting for one is a mistake. The practical approach is to measure what you can and model the rest, as long as everyone agrees on how the modelling works.

Ownership is the harder problem. Daniel's advice was to map every stage from marketing spend to closed sales revenue, find where revenue leaks, and give each stage an owner. Shamsul's version was a triangle with customer experience at the top and marketing and sales as the base, each with clear and equal responsibility.

Six practical takeaways for responding to the AI search shift

• Track AI referrals as their own channel. Add chatgpt.com, perplexity.ai and gemini.google.com to your referrer reporting, and give AI its own place in your channel taxonomy so it isn't misattributed to PPC or direct.
• Measure calls, not just clicks, and feed outcomes back into bidding. Report visit-to-call rate and high-intent calls alongside traffic. As Daniel said, the algorithm can't bid against value it doesn't see, so extract the intent and outcome of each call and pass it to your bidding algorithms.
• Fix the close, not just the traffic. Brief sales and call-handling teams on price and budget objections, and move scripts from education to closing. Callers have often done the research already.
• Turn call conversations into content and creative. Use what callers actually ask to shape website FAQs, messaging, creative and retargeting, so the answer is there before the question is asked.
• Map the journey and name an owner for every stage. From media spend to closed revenue, know where you lose buyers and who is responsible for fixing it.
• Build for AI discovery the long way. There's no silver bullet. Invest in content depth, review volume and structured data, and focus on being relevant and authoritative for the questions your buyers ask.

The AI search shift restructures the buyer journey – it doesn't end it

Discovery and comparison are moving earlier in the journey, but the moment of commitment still matters enormously, and for many considered purchases it happens on a call. The brands that win won't just chase clicks. They'll understand where intent forms, where conversions break down, and use those signals to improve bidding, messaging and the sales conversation. Or, as Shamsul put it:

"If you don't like change, you will like obsolescence even less."