The SEO-GEO Gap

Something strange happens once you start segmenting your analytics by referral source. Google rankings hold steady, click-through rates look healthy, and yet a growing share of visitors now arrive through ChatGPT, Perplexity, or another AI assistant rather than a conventional results page. These two visitor groups do not behave the same way, and more importantly, they do not respond to the same content.

A recent study from Search Engine Land examined 150,000 pages across ten websites over a single month. The findings reveal a pattern that content teams cannot afford to ignore: what performs on Google and what gets cited by AI are often two separate outcomes, sometimes overlapping, often not.

AI Search and Google Search Reward Different Content

Here is the core finding. Content that ranks well on Google does not automatically get surfaced or quoted by AI tools, and the gap between the two is wider than most marketers assume.

Posts built around original data, emerging trends, or independent analysis earned citations from AI tools between 61% and 78% of the time. Generic how-to guides and standard educational articles, by contrast, were cited only 12% of the time.

Consider why that split exists. Large language models can already produce a competent "how to do X" explainer without any outside help, so there is little incentive to quote a page that says the same thing everyone else does. A page becomes citation-worthy the moment it contains something the model cannot generate on its own: a proprietary statistic, an original argument, or a finding published nowhere else.

The implication is direct. Give AI tools something they cannot reproduce internally, and they will reference it. Give them a rehashed explainer, and they will write around you instead.

Your Best Google Pages Are Probably Not Your Best AI Pages

One statistic from the study deserves particular attention: among the top 100 organic search pages analyzed, 49 received zero traffic from AI referrals. That is not a rounding error or a handful of outliers. Nearly half of the pages driving the strongest Google performance contributed nothing to AI-referred traffic.

The imbalance shows up again at the top of the funnel. The ten highest-performing organic pages captured 55% of all Google traffic, yet accounted for only 29% of AI referral traffic. Two genuinely different audiences are moving through two genuinely different channels, and a page built to dominate one channel offers no guarantee of success in the other.

Content strategy now needs to account for both realities simultaneously: pages engineered to rank on Google, and pages engineered to be quoted inside an AI-generated answer. Treating them as the same objective risks optimizing for only half the opportunity.

Product and Service Pages Outperform in AI Traffic

This result surprises most people who assume AI referral traffic mirrors organic search patterns. When the study measured AI sessions relative to overall page traffic, service and product pages outperformed every other content category.

When a user asks an AI assistant a question like "what tool should I use for X" or "who does Y best," the response frequently links straight to a service page rather than a blog post. That behavior turns product and service pages into a genuine acquisition channel, one that deserves deliberate optimization rather than being treated as an afterthought to a blogging strategy.

What AI Users Actually Do on Your Site

Average session duration looks nearly identical across the two groups: roughly 47 seconds for both Google visitors and AI-referred visitors. That average, however, conceals a meaningful behavioral split underneath it.

AI-referred visitors spend less time on articles, and the likely explanation is that they arrive scanning for one specific piece of information rather than reading for context. Yet on tools, homepages, and product pages, the same visitors linger two to three times longer than their Google-referred counterparts.

That contrast says something useful about intent. AI users tend to arrive with a specific task already in mind, shaped by the question they asked the assistant. A page that helps them finish that task keeps them engaged. A page offering only a broad overview loses them almost immediately.

Three Things Worth Acting On

Lead with a direct answer. Any page you want an AI tool to cite should open with a clear, concise response to the primary question it addresses. AI systems scan for the most extractable answer on a page, so front-loading that answer, rather than burying it under introductory throat-clearing, determines whether your content gets quoted.

Make original data the headline, not a footnote. Proprietary research, survey results, customer data, or any insight unique to your company represents the strongest asset available for AI visibility. Push it to the top of the page instead of leaving it for paragraph six.

Do not overlook your interactive tools. Calculators, assessments, configurators, and quizzes attract AI referrals at a disproportionately high rate relative to their traffic volume. Name them clearly, describe their function in plain language, and confirm they remain accessible to crawlers as well as human visitors.

The Bigger Picture

Generative Engine Optimization or GEO, is not a replacement for SEO. It runs alongside it as a parallel discipline. Google evaluates content against one set of criteria; AI tools apply a different lens entirely, weighing originality and extractability over the signals traditional search has trained marketers to chase for two decades.

A strong organic strategy still carries weight, but it no longer covers the full scope of how people find and consume information. The websites performing best across both channels in this study shared one trait: they built content that answers specific questions with specific, original information, useful to a person clicking through from Google and equally useful when an AI assistant hands that same content to a user as the answer itself.