Skip to main content

The context

AI-referred traffic converts at two to three times the rate of traditional search. It spiked 54% higher on Thanksgiving and 38% higher on Black Friday. And it accounts for 0.13% of total web sessions. That tension — elite conversion, invisible share — is the defining asymmetry in marketing right now. Meanwhile, Semrush reports an 800% year-over-year increase in referrals from large language models, while traditional search volume is predicted to decline 25% by 2026. Generative Engine Optimization (GEO) is not a complement to SEO. It is replacing the traffic logic underneath it.

What I tried / what I saw

The AI traffic conversion data (ALM Corp, 2 million LLM sessions):
“AI-referred traffic converts at 2–3x the rate of traditional search — with holiday spikes of 54% higher on Thanksgiving and 38% higher on Black Friday. Yet AI-powered search still represents only 0.13% of total web traffic (1 in 769 sessions).”
The citation ecosystem finding:
“In consumer and financial categories, 65%+ of the sources LLMs reference are publishers, user-generated content, and affiliates — not owned properties. A brand’s own website is a minority signal in how AI forms recommendations.”
Your site is 5–10% of what an LLM cites when recommending your category. The other 90–95% lives in editorial coverage, Reddit threads, affiliate reviews, and forum discussions you do not control. The “Share of Model” gap:
“The emerging metric: how often does an AI recommend your brand? Called ‘Share of Model’ — replacing Share of Voice, SOV, and traditional search share. Only 13% of organizations have embedded agentic AI for brand discovery and search.”
The Walmart counterdata:
“Walmart’s ChatGPT-powered checkout converted 3x worse than the traditional website experience. Shannon Millard (Epsilon): ‘consumers will always want the freedom to explore, compare, and take pleasure in discovery.’”
AI-powered checkout converting at one-third the rate of the standard funnel signals that the browsing and discovery phase is not something consumers want automated away.

What sticks

The conversion rate advantage of AI-referred traffic is real and proven. The traffic share is 0.13% — which means the window to establish position before the channel gets crowded is still open. Three things to act on now:
  1. Stop optimizing only for your owned site. Your website is a minority signal. The citation ecosystem — publishers, UGC, affiliates — is where AI forms its recommendations about your category. Influence that.
  2. Start measuring Share of Model, not just Share of Voice. How often does an AI recommend your brand? At 13% adoption of any embedded strategy, the field is effectively uncontested.
  3. Don’t confuse AI-driven discovery with agentic checkout. The asymmetry to capture is AI-driven discovery routing high-intent users to experiences they control — not automating the purchase decision itself. The Walmart data makes this distinction concrete.
The brands that build Share of Model before the channel scales are buying the equivalent of search rankings before Google was the default. The competition has not arrived yet.