Is AI search a distraction? Not for hotels, judging by our data

The AI visibility debate

Between 14 and 22 October 2026, three pieces of research on AI visibility were published or updated, all asking how much AI search matters and how far it can be tracked. The SEO Works published a twelve-month benchmark of AI referral traffic across 95 websites. Kevin Indig went looking for AI traffic hidden inside direct visits. Marie Haynes looked at twelve months of Google Analytics data across 100 sites and asked whether GEO is a distraction.

Rand Fishkin walked through all three in a short video and came away less sure that visibility in ChatGPT and other AI assistants makes much business difference for most companies, although he set Google’s own AI Overviews and AI Mode aside as a separate question. Mike King answered with a post arguing that AI search should be judged as a brand channel rather than a traffic channel and Gianluca Fiorelli argued that AI visibility can’t be judged on referrals and conversions alone. LinkedIn predictably exploded in discussions around the data and the conclusions.

Marie, Kevin, Rand, Mike and Gianluca are all established researchers and commentators in the wider SEO and AI search industry, whose work I’ve followed and valued for many years. But while reading the studies, I kept thinking that the data and some of the conclusions didn’t quite match what we are seeing in hospitality.

Is hospitality different?

None of the studies as far as I know focuses on a particular industry, and none publishes a hospitality-specific breakdown. Our dataset is smaller, but it is entirely hospitality and travel.

What we’re seeing suggests that AI assistants may already play a bigger role in hotel discovery than the wider industry figures would lead us to believe.

This post isn’t intended to set any type of benchmark. It’s a sector-specific comparison to the wider studies, and we’ve tried to be clear about the baseline behind every figure: organic search, all visits, or the same months a year earlier. Like the other studies, we’re looking at referrals from AI assistants such as ChatGPT using Google Analytics source data.

Our hospitality data

We looked at the hotels and resorts we manage, which are mostly boutique properties (most get fewer than 10,000 visits a month), and separately at hospitality-adjacent businesses (not hotels but in the travel sector).

Between September 2025 and August 2026, the same window The SEO Works used, the hotels had about 3.1 million sessions, around 30,000 of them from AI assistants. The travel businesses had about 660,000 sessions, around 17,000 of them from AI assistants.

In total, we looked at 37 websites with about 3.8 million sessions, around 47,000 of them from AI assistants. Where we could, we measured things the same way the studies did. All figures in the table below cover September 2025 to August 2026.

For hotels, the AI share is much higher than the averages in the other studies. Against Google organic sessions, our hotels see roughly eight times the share in Marie’s dataset, and ChatGPT alone is close to nine times her ChatGPT figure. Against all organic search, our hotels see roughly one and a half times The SEO Works’s figure, and more than double it by August 2026. The SEO Works found organic search sending 39 visits for every AI visit; for our hotels it was 24, and for the travel businesses 14.

As a share of organic search, AI visits to our hotels went from 1.4% in September 2025 to 4.3% in April 2026 and 9.2% in August 2026, while The SEO Works’s figures for the same months were 2.25%, 2.14% and 3.86%. For the travel businesses, the same months were 2.8%, 6.3% and 10.3%.

We have some reservations about using organic search as a baseline to compare Ai visits with. Most of the hotels in our dataset run paid search campaigns, and many of them have brand protection campaigns running also. Some users who would have arrived through organic search arrive through the ad instead. So a hotel’s AI share of visits versus organic search moves partly with its paid search strategy, which is why we also show AI against all visits.

AI Assistant visits as a share of organic search visits

AI share is growing, and not only because organic is shrinking. A rising share of organic can mean AI is growing, organic is falling, or both. Across the same hotels in both years, both are happening, but most of the change comes from AI growth.

From January to September 2026, AI visits were about six times the same months of 2025, and every single month was at least four and a half times its 2025 equivalent, and up to thirteen times. Over the same months, Google organic visits fell by about a third and total visits were flat.

For the travel businesses, AI visits roughly tripled while Google organic fell by about a fifth. Branded search campaigns matter here. When paid brand traffic replaces some organic visits, a decline in organic sessions doesn’t necessarily mean a decline in search demand or SEO performance. Keep that in mind when comparing the organic search figures.

Year-on-year growth in AI visits is something almost every site will likely show by now, so we also looked at AI visits as a share of all visits. This gives us another way to assess growth without relying on organic search as the baseline. On the same hotels, the AI assisted share rose in every quarter, from under 0.1% of all visits in the first quarter of 2025 to about 1.9% in the third quarter of 2026.

For the travel businesses it rose in every quarter to the third quarter of 2026, from about 0.3% to 3.7%, with a dip in September which may be partly season related. Marie found ChatGPT referrals across her sites falling from August 2025. Ours rose over the same period, with a sharp step up from May 2026 when ChatGPT started linking brand mentions more often.

AI Assistant visits as a share of all website visits

It is not a DuckDuckGo-sized channel. Rand pointed out that the AI share in Marie’s data is smaller than DuckDuckGo’s. On our hotels, AI assistants sent about four and a half times as many visits as DuckDuckGo over the twelve months, and between June and August 2026 they sent more than Bing.

Booking intent

About three in four AI visits to our hotels landed on the homepage, which fits a guest who has been given the hotel’s name and clicks through. These visits engaged more than organic visitors (75% of sessions engaged, against 69%).

The clearest intent signal is what they did next. Many of our hotel setups track clicks into the booking engine. It shows that the user arriving at the website was interested enough to click through to the booking engine to check pricing and availability. Across the months that tracking was live between September 2025 and August 2026, covering about 21,100 AI visits, the share of visits that clicked through to the booking engine was:

These are pooled rates across those hotels, using each hotel’s own booking-engine click event for every channel. Hotel by hotel, AI visitors clicked through more often than the site average at all but one hotel (a median of about twice the average), and more often than organic visitors at every one of them (between about 1.1 and 2 times, a median of 1.5).

Here again we choose to measure this intent against the site average as well as organic search for a reason. Organic search has been getting harder to measure in Google Analytics and may be influenced by marketing campaigns. Google organic visits to our hotels fell by about a third over the past year while total visits held steady. Comparing AI visitors with every visitor to the site is less dependent on how well or in what mix organic search is being counted in Google Analytics.

The travel businesses in our data, which mostly take enquiries rather than online bookings, look much more like the studies’ conversion findings. AI visitors there sent an enquiry or booked at nearly three times the Google organic rate, and more often than organic visitors at every site with enough AI traffic to compare.

Why don’t recorded bookings follow?

Recorded hotel bookings do not mirror the intent trends we see above. On our hotels, AI visits converted to a recorded booking at about two thirds of the rate of all organic search visitors, and about four fifths of the Google organic rate shown in the table (visitors from other search engines book more often than Google’s).

Against the site average the recorded rate for our hotels is lower still, at about 0.6 times, though this is the comparison most affected by how some booking engines pass the visit source back to Analytics. Leaving out the hotels where recorded bookings lose their original source, and comparing the remaining hotels’ AI visitors with their own site average, it rises to about 0.9 times. The SEO Works found AI visitors converting at 1.37 times their site average, and our travel businesses, at 2.2 times, sit above that.

The difference for hotels may partly be what we count: The SEO Works and Marie measured key events, which may include enquiries and other actions, while for hotels we measured only completed bookings.

We cannot say with 100% certainty why the hotel booking trends diverge from the intent trends. Using Google Analytics to attribute bookings and other commercial results, in any vertical, has become a minefield of caveats and parentheses. Our thoughts on it centre around the guest journey from hotel discovery to booking and how that journey is recorded in Google Analytics itself:

• Hotel bookings happen on a separate booking engine, and tracking across the two domains varies.
• Cookie consent and Google consent modes can alter or cause data loss.
• Privacy features on guests’ devices may drop indicators of the original source.
• Guests switching devices between discovery and booking complicates attribution.
• A guest who discovers a hotel through ChatGPT may come back through a Google brand search or ad click before booking.

My own view, which this data cannot prove by itself, is that AI assistants are becoming an important hospitality discovery surface: a guest asks for recommendations, gets a shortlist, explores the hotel websites and pricing, then comes back later, often through a brand search or a direct visit, and completes their booking.

Google becomes the brand doorway and takes the credit. That fits Mike King’s case for treating AI as a brand channel, and the Similarweb research Kevin Indig cites, where people given AI brand recommendations were more likely to visit those brands in the following week, mostly through search. It would also help explain why AI visitors reach the booking engine so often while recorded bookings lag.

Gianluca Fiorelli sketches almost exactly this journey for a family holiday: the AI conversation shapes the choice of destination and hotel, but may never appear in the booking’s attribution.

A hospitality twist: AI visits that come through your listings

A twist which may also be more specific to hotel visibility than wider industry trends has been apparent in our data for many months now. Most of our hotels tag the website link on their Google Business Profile, and some tag their Yelp link too, so we can see those visits in Google Analytics.

Between June and September 2026, on the hotels that tag their Business Profile link, nearly one ChatGPT visit in four still carried the Business Profile or Yelp tag as the session medium. On three hotels it was over 40%, and on two it was over half. On the travel businesses that tag their listings, it was nearly two in five.

Those tags only exist on the hotels’ Google Maps or Yelp listings, which is strong evidence that the links ChatGPT shows for these hotels come from those listings, directly or via the data built on them.

If the website link on your Google Business Profile and your other listings is not tagged, tag it. Without a tag, these visits look like any other ChatGPT click.

Google’s AI channel misses some visits

Since mid May 2026, Google Analytics has had a default AI Assistant channel. Under Google’s definition, a visit only goes into it when its medium is exactly ai-assistant, a value Analytics assigns itself when it recognises an AI referrer. In our data, visits that already carry a tracking tag, such as those listing links, keep their own medium and mostly land in Unassigned.

From June to September 2026, the first full months after the channel launched, Google’s channel caught 80% of unpaid AI visits to our hotels and 61% for the travel businesses. If you rely on it alone to assess your website’s AI traffic, you will likely be undercounting.

What this may indicate

The wider studies have helped put some perspective on the increasingly outsized discussion around AI visibility. Our hospitality data adds another perspective: AI referrals may be more significant in some sectors than the broader averages suggest.

In our hotels, AI visitors reach the booking engine at nearly twice the rate of organic search visitors and more than two and a half times the site average. Recorded bookings don’t follow the same pattern, but that doesn’t make the behaviour any less interesting.

Hospitality is still centred on a human experience. Particularly for boutique hotels, guests may continue researching long after an AI recommendation, comparing options and returning through other channels before making a decision.

For hotels, our practical reading is not to divert budget from SEO into a separate GEO discipline, but to treat AI assistants as another discovery surface built on the same foundations. What our data adds is that, in the hospitality businesses we work with, that surface already sends a meaningful and growing share of visitors, and those visitors show strong booking intent.

A call to the hospitality industry

Our numbers are small next to the previously mentioned studies, but the differences seemed striking enough to share.

Two things to keep in mind. Most of the hotels in this study are boutique (in other words, not chain hotels or motels and not business or conference focused). They target experience-led travellers, not simply guests looking for a place to sleep for the night. Many of them also work with us on AI visibility. That is not a random sample of hospitality, and it may show more AI activity than an ‘average’ hotel would. And our data shows what AI visitors do, not why: a high booking-engine click rate fits AI introducing qualified potential guests to hotels, but it does not prove it.

I’d be fascinated to see what other hospitality businesses and agencies have to say. If enough of us share what we’re seeing, there will be more data to judge whether AI visibility in hospitality really does separate itself from wider trends, or whether our data set is an outlier in some way.

Check your own site: open our free Data Studio AI visits report while signed in to Google, make a copy, connect it to your own Google Analytics 4 property, and let us know what you see.