Vitaliy Levit killed Gondola’s six-figure SEO services line last year because he could no longer promise it would work. That is the strongest signal in the room about how completely AI has rewritten search for tour operators.
Vitaliy joins Aubrie for an exclusive preview of five months of research into how AI search is reshaping visibility for tour and activity operators. Gondola builds and manages hundreds of operator websites with a normalized technical architecture, which gives Vitaliy a controlled data set most agencies cannot produce. Across 35 sample sites and 15 prompts each spanning location, offer, and personalized queries, his team tracks citations across ChatGPT, Gemini, Perplexity, Google AI overviews, and Google AI mode. The numbers explain what operators have been feeling on every call: aggregate Google search traffic to their sites is down 36 to 38 percent year over year.
The webinar covers which backlinks correlate with AI citation and which do not, why directory and aggregator links no longer pay off, why publishing more comprehensive information beats publishing pretty pages, and the specific experiment that lifted AI visibility 44 percent on the test sites. Vitaliy also explains where individual operators can actually compete (offer-specific and personalized prompts), where they cannot (location-specific prompts, where OTAs and Google Maps dominate), and why Google Business Profile optimization is now the single highest-leverage move for location visibility.
Resource offered to attendees: Vitaliy shared a Typeform where operators can submit their Google Business name and receive back a CSV of their reviews plus the prompt his team uses to generate the FAQs for the LSH page: https://form.typeform.com/to/Fu5K4YIH
Takeaways
- Operators with stable, decade-old businesses are seeing direct bookings drop 30 to 40 percent without changing anything. 03:23 – 04:30 The first signal most operators pick up is a booking problem they cannot explain. Vitaliy hears this on weekly calls from people who have not changed pricing, tours, or site structure. The cause is a shift in how consumers find and choose tours, not anything specific to their business.
- Ranking number one in Google organic results no longer puts you above the fold. 05:50 – 06:30 AI overviews, ads, common questions, and the maps pack stack above the top organic result, pushing it to roughly position fifteen. The 50 to 60 percent click share that used to belong to the top result is now distributed across surfaces that mostly do not drive an outbound click.
- AI overviews have removed 50 to 60 percent of all clicks that used to leave Google. 06:21 – 07:30 Google now answers most queries directly inside the AI overview. The high-funnel traffic that newsletter pop-ups, retargeting audiences, and content marketing depended on is gone. Playbooks built on top-of-funnel visitors landing on a website have stopped working.
- Search traffic across hundreds of operator sites is down 36 to 38 percent year over year, while LLM traffic is rising from a tiny base. 17:00 – 18:30 A single operator site averages roughly eight to twelve LLM visits per month versus 1,500 to 2,000 Google visits. The LLM channel matters, but it will not replace Google search in absolute volume any time soon. Both numbers move in opposite directions, but the scale gap is real.
- Branded search is up. Local discovery searches are down 48 percent year over year. 18:59 – 20:30 Across the Gondola network, brand-name queries have increased, meaning people who already know you still find you. Brand-plus-activity queries are down 17 percent. Local discovery queries like “things to do in [city]” are down 48 percent because that surface is now owned by AI overviews, the maps pack, and Google’s things-to-do panel.
- The total number of backlinks pointing at your site has no correlation with AI citation rates. 24:10 – 24:39 Directory links, link-building services, and offshore link placements produce no AI visibility benefit. Volume by itself does nothing. Operators who paid for SEO link campaigns over the past few years should expect zero AEO carryover.
- Domain rating has only a slight correlation with AI citation and is not statistically significant. 24:39 – 25:30 DR was the metric SEO professionals built their reporting around for years. It does not predict AI visibility. A high DR built through generic link-building does not translate to LLMs choosing to cite you.
- Vitaliy killed Gondola’s six-figure SEO services product line in 2025 because he could no longer guarantee it would work. 31:10 – 32:50 The foundational SEO package was profitable and reliable. Operators paid roughly $1,200 to $2,000 and could expect performance gains within months. With the AI overview rollout, Vitaliy stopped offering the service because he was no longer confident the work would produce results. SEO is now table stakes, not a growth lever.
- AI does not browse your website. It pattern-matches against the information you publish. 35:00 – 35:30 The last decade taught operators to design for human navigation: short copy, clear hierarchy, fewer words, more visual impact. AI agents work the opposite way. They want comprehensive structured information that matches against what they already know about a user. The mental shift is from “make it pretty for humans” to “make it dense and machine-readable for agents.”
- Reddit is still cited by LLMs but at lower weight than it once was, because marketers tried to game it. 36:35 – 37:00 Reddit was an early strong signal for LLM citation. As soon as that was discovered, marketers flooded the platform with promotional content, and the LLMs downweighted it. Reddit still appears in citations, just far less than its share of total discussion would suggest.
- First-person versus third-person voice on a page has no correlation with AI citation. 37:00 – 37:30 Vitaliy expected pages written from a personal point of view (“I went on this tour and…”) to outperform third-person articles. They did not. Authored voice does not give AI a stronger signal than well-structured third-person content.
- Page length and word count on a single page do not predict AI visibility. 37:30 – 38:00 A 3,000-word page does not get cited more than a 500-word page if the content quality and structure are equivalent. What matters is information density and proper structuring, not raw word count.
- On-topic backlinks from authoritative sites with descriptive anchor text are what drives AI citation. 37:30 – 39:30 If a publication writes a dedicated article about your business and links to you multiple times with anchor text like “Pagosa Outside, a guided river rafting outfitter in Pagosa,” that single citation outweighs hundreds of generic directory links. The link has to live inside on-topic content, on an authoritative domain, with anchor text that describes you specifically.
- Travel publications, DMOs, tourism boards, news outlets, educational sites, and independent travel blogs are the sources AI pulls citations from. 39:56 – 40:50 The top-cited operators in the study have links from these categories. The bottom-cited operators have most of their links from directories and aggregators like TripAdvisor and Viator. Five strong publication links will beat 500 directory listings.
- You can run your own outreach to local DMOs, regional publications, and travel bloggers using ChatGPT or Claude. 43:15 – 44:24 A professional PR firm is fine if you have the budget, but it is not required. Ask the AI to find local DMOs and travel publications in your area and draft outreach for an event you are hosting or a story angle you can offer. The barrier is time, not capability, and the right outreach to five strong outlets is achievable.
- The LLM Start Here page lifted AI visibility 44 percent across the five tested models in five months. 45:55 – 50:30 This is one page on your site containing every piece of information an LLM might need: product details, every category, every page summary, every customer review, and 300 AI-generated FAQs in proper Q&A schema. Vitaliy ran the experiment across his test sites and saw citation lift on Google AI mode, Gemini, AI overviews, Perplexity, and ChatGPT.
- The reviews already on your site are the most underused asset for AI visibility. 48:23 – 49:30 Gondola pushed one thousand customer reviews into a single page in plain text, then ran the review and product data through Gemini’s one-million-token context window to generate 300 FAQs in proper schema. Reviews contain the exact language real customers use to describe the experience. LLMs ingest that language and use it to match new prompts to your business.
- For location-specific prompts, individual operators almost never win citations. OTAs, DMOs, and Google Maps dominate. 53:17 – 54:30 “Best things to do in Grand Cayman” is cited overwhelmingly by TripAdvisor, Viator, Yelp, Discover Puerto Rico, Visit Anaheim, Wikipedia, and Get Your Guide. Operators trying to win these high-funnel queries are competing against entities ranked specifically for that purpose. Optimize for offer-specific and personalized prompts instead, where individual operators can actually compete.
- Google Business Profile is now getting cited heavily in Gemini and Google’s AI surfaces for location queries. 54:30 – 55:10 Higher Google Maps ranking translates directly to more Gemini and Google AI traffic. This is the single highest-leverage optimization for location-specific visibility because operators can actually influence their map ranking through reviews, posts, and category accuracy. Whatever attention has been going elsewhere should go here first.
- Rich schema and structured data are even more important now than they were for SEO. 01:08:58 – 01:09:45 A well-built homepage has 25 or more valid rich data structures: product snippets, merchant listings, carousels, FAQ, local business, organization, reviews. AI crawlers consume this data preferentially because it is pre-tagged and unambiguous. Sites with weak or inconsistent structured data are invisible to the systems now driving a growing share of bookings.