Improving the landing experience is worthwhile, but not because “AI traffic is about to replace search.” The goal is to make it easier for visitors who already know what they need to find products and take the next step. Problems with homepage navigation and on-site search also affect visitors from non-AI sources. Investment decisions should therefore not rest solely on AI’s share of traffic, nor should high conversion rates alone justify a major increase in spending. A sounder approach is to first establish where visitors actually land and how they interact, then prioritize removing obstacles that affect visitors from multiple sources.
Visitors’ needs still matter when traffic is low
In a study of roughly 35,000 websites published in March 2025, Ahrefs found that AI accounted for only about 0.1% of total referral traffic. That was the result for the sample at the time, not a fixed proportion for every website today. The article also noted that analytics tools may record some AI traffic as direct traffic. Even without tracking issues, however, clicks may remain limited because users can complete their research within AI platforms without visiting a website.
Improving the experience need not depend on a surge in traffic. An August 2026 article in Search Engine Journal argued that helping visitors with clear needs find products faster through the homepage is valuable in its own right, as is providing on-site search results that match their intent. AI referrals expose existing problems rather than create entirely new needs.
At Procrustes, we handle page planning, design, and functionality for partner websites, alongside content planning, writing, and updates.
Check where visitors land—do not assume it is the homepage
A 2026 Similarweb report summarized in the Search Engine Journal article estimated that 58.8% of ChatGPT referral traffic landed on homepages. Meanwhile, a Previsible analysis covering 166 websites and 6.77 million AI-referred sessions found that 28.8% of ChatGPT referrals landed on on-site search results pages. These findings suggest where to look, but cannot be applied directly to every website.
By contrast, a client study published by Seer in September 2025 found that only about 7% of AI visits landed on homepages, with most reaching deeper pages. Because the study periods, samples, and referral sources differ, these findings cannot establish where visitors generally land across websites.
- If most visitors land on the homepage, check whether people with clear needs can find relevant products.
- If they land on on-site search pages, check whether the results meet their needs so they do not have to start searching all over again.
- If traffic is concentrated on informational pages, examine engaged sessions by page and purchase stage, rather than tracking only total visits.
Assess traffic share, growth, and conversion together
Seer’s 2025 study recommended tracking traffic share, growth, and conversion rates together: focusing only on conversion rates may lead to overinvestment, while focusing only on traffic share may cause opportunities to be overlooked. Conversion rates help explain the value of traffic, but they do not capture its full impact. Improving a page is also no guarantee of increased conversions.
In another article in 2026, Seer cautioned that rising session counts can coincide with falling engagement rates and traffic concentrated on a single glossary page. A sounder approach is therefore to remove confirmed obstacles in the landing experience first, then use page performance to decide whether to expand investment, rather than undertake a wholesale redesign based solely on claims of AI traffic growth.
Improving the experience for AI-referred visitors: perspectives from across the web
Do AI referrals reported by analytics tools capture the full impact?
Not necessarily. Ahrefs noted in 2025 that analytics tools may classify some AI referrals as direct traffic, so reports may identify fewer AI-referred visits than actually occurred. However, even without tracking issues, users may do more research within AI platforms without clicking through to a website. Traffic reports are worth checking, but that does not justify assuming all visits from unidentified sources come from AI.
Are small numbers of AI visitors necessarily more likely to convert?
There is no universal answer. In the Cardmarket case, AI referrals remained a small share of traffic over the 12 months described in the article, but their conversion rate was substantially higher than that of other sources. That was the result for that platform, not a guarantee that other websites will see the same outcome. Seer also noted that conversion rates vary widely across industries and should be assessed alongside traffic share and growth, rather than using a single high conversion rate to justify increased investment.
Can rapid AI referral growth be directly extrapolated to determine next year’s investment?
Direct extrapolation is not advisable. AI referral traffic to Seer’s clients grew by 113% from May through July 2025 compared with the preceding three months. The author also noted that the year-over-year increase was affected by a low baseline. The future growth multiples in the article were scenarios based on different quarterly growth rates, not forecasts. When assessing investment, recent growth should be treated as a signal, not a promise.
Are the pages cited by AI the same pages visitors actually land on?
Not necessarily. Similarweb’s 2026 report estimated that 65% of the URLs cited by ChatGPT were at the second or third level of a website’s structure. Separately, 58.8% of ChatGPT referral traffic landed on homepages. These measure different things and cannot be used interchangeably. The figures are sample-based estimates and do not mean every website follows the same distribution.
References
- AI Makes Up 0.1% of Traffic, but Clicks Aren’t Everythingahrefs.com · 2025-03-26T09:42:22+00:00 — Discusses: Do AI referrals reported by analytics tools capture the full impact?
- AI Visibility Is Lying to You: There’s No One-Size-Fits-All AI Search KPIwww.seerinteractive.com · 2026-08-31T15:59:48-05:00 — Source used in the article above.
- AI Referrals Are Recreating The Oldest Mistake In Conversion Optimizationwww.searchenginejournal.com · 2026-08-05T20:12:01+00:00 — Discusses: Are the pages cited by AI the same pages visitors actually land on?
- How Cardmarket wins search by knowing its community better than anyonewww.semrush.com · 2026-07-07T13:00:00Z — Discusses: Are small numbers of AI visitors necessarily more likely to convert?
- Study: The AI Search Landscape, Beyond the SEO VS GEO Hypewww.seerinteractive.com · 2025-09-22T19:16:38-05:00 — Discusses: Are small numbers of AI visitors necessarily more likely to convert?; Can rapid AI referral growth be directly extrapolated to determine next year’s investment?
