Your booking attribution is about to get worse. AI search is the reason.

TL;DR

• Martin Soler (partner at Soler & Associates, a hospitality marketing practice) says zero-click artificial intelligence (AI) search is thinning the trail from discovery to booking, and 68% of US Google searches already end without a click.
• Chrome kept third-party cookies, so the gap appears to come from guests who research an independent hotel in AI answers and never visit its site.
• Independent hoteliers can ask guests how they found the property, add an AI channel in Google Analytics 4 (GA4), and compare direct bookings with website sessions this quarter.

Soler argues in a Hospitality Net (a B2B hotel industry news site) opinion piece that hotel attribution was always closer to guesswork than science and is about to get worse. AI answers questions without sending users to websites, so the path from discovery to booking fades.

Guests still need somewhere to sleep, Soler says, so demand remains. His advice is to ask guests how they heard about the property. The data shows how fast.

Zero-click searches now reach 68% of US Google searches

SparkToro (an audience research software company) analyzed US Google searches from January to April 2026 with data from Similarweb (a web analytics company), Search Engine Land (a search marketing news site) reports:

• 68.01% ended without a click, up from 60.45% in 2024.
• AI Overviews appeared on more than 20% of searches and cut click-through by nearly 60% when they did.

The study covers all US Google searches, not hotel searches.

One in three US travelers research trips on AI platforms, five times the rate of two years ago

Phocuswright (a travel industry research firm) found:

• 33% of US travelers use generative AI platforms such as ChatGPT for trip research.
• 44% would book directly inside an AI platform.

Chrome kept its cookies, which leaves AI answers as the new blind spot

Google dropped its cookie phase-out in 2024, confirmed in April 2025 that Chrome would keep its approach, and retired most Privacy Sandbox alternatives on October 17, 2025, per Stape (a server-side tracking provider).

AI answers leave a partial trace in GA4:

• ChatGPT desktop citation links have carried utm_source=chatgpt.com since mid-2025, but not on every link, per Mo Agency (a digital marketing agency).
• GA4 often files those visits as Referral or Unassigned, and clicks from Google's AI Overviews and AI Mode count as Organic Search, per Themeisle (a WordPress resource site).
• A guest who reads an AI answer and never clicks leaves no trace.

The AI conversation that starts a booking likely never appears in analytics

A booking path that starts in a chatbot

Picture a parent with the kids off school and four days of leave who asks an AI assistant: "Where can I take my kids for four days? They like beaches, pools, and theme parks." The assistant suggests places to stay.

The parent might then follow the assistant's link to a hotel's site, or open its Instagram page after the assistant cites it, and book. BrightEdge (a search engine optimization software company) found ChatGPT cites Instagram mostly through profile pages.

The Direct channel collects credit it did not earn

Analytics record a branded search, an Instagram visit, or a Direct visit, and the AI conversation never appears. Work that puts an independent hotel into AI answers then appears to do nothing, while the Direct channel likely collects credit for demand the AI created.

Independent hoteliers can recover part of the AI step with three checks

• Ask every guest how they found the property. Guest answers capture the AI step analytics miss. Add the question to the booking confirmation, list ChatGPT and Google's AI answers as separate options, and compare replies with GA4's Direct channel after 90 days.
• Add an AI channel in GA4. AI referrals otherwise land in Referral or Unassigned. Build a custom channel group for chatgpt.com, perplexity.ai, and gemini.google.com plus the utm_source tag, rank it above Referral, and review it monthly; it will undercount.
• Compare direct bookings with website sessions. Sessions falling while direct bookings hold likely means guests find the property elsewhere and return to book. Chart both monthly for 12 months, and read a joint fall as weaker demand.

The dashboard will keep getting worse, but the guest can still tell you where the booking started.