AI recommendations trusted more than traditional search by 70% of travellers, study finds

Some 70% of travellers trust AI recommendations more than traditional search results, according to a new behavioural study by travel marketing agency Propellic.

The State of AI Visibility for Travel report examined how travellers use artificial intelligence (AI) to research and plan trips, finding that generative AI is emerging as a new competitor to online travel agencies (OTAs) by directing users towards travel operators' direct booking channels.

The study found that 71% of participants were directed to operators' own websites when moving from AI recommendations to booking, while 68% said they preferred to book directly.

However, the research also highlighted friction in the transition from AI-powered trip planning to completing a booking, with more than three-quarters of booking handoffs failing to retain key information such as travel dates and guest numbers.

The research was conducted with Touchstone Research in August 2026 and followed participants over a week as they used their preferred AI tools to choose destinations, research hotels, build multi-day itineraries, find excursions and progress towards booking.

Unlike traditional surveys, the study recorded participants' screens as they planned trips, rather than asking them to rate their perceptions of AI.

It is Propellic's second study into AI's impact on travel, following its 2025 From Search to Purchase report, which focused on how Google AI Mode was changing the booking journey.

Participants rated AI recommendations 4.1 out of five for accuracy, with only 5% saying they did not trust recommendations from AI tools.

ChatGPT accounted for 47% of AI tool use, followed by Google Gemini at 22% and Google AI Mode at 20%. Most participants largely remained within a single AI tool throughout the week, rarely checking a second platform.

The findings suggest travel brands may need to rethink how they approach visibility, as travellers increasingly form their preferences through AI-generated recommendations before visiting a travel company's website.

Brennen Bliss, CEO and founder of Propellic, said: “Most AI research in travel asks people what they think. We watched what they did, and the shortlist of travel preferences was selected before anyone reached a website.

“The work for every travel brand is the same: be in the answer, and format your content strategy so that a machine can quote you.”

The report identified eight key findings, including the importance of structured content, reviews and imagery in influencing AI recommendations.

Recognisable travel brands featured in AI-generated answers were sometimes perceived by participants as paid placements, even when the recommendations were organic. By contrast, named local operators with contact details were more likely to be interpreted as genuine recommendations.

The research also found that comparison tables and ranked lists helped build trust in AI recommendations, while reviews and photographs were the most-clicked elements of hotel listings.

When asked to create multi-day itineraries, participants planned trips themselves rather than looking for existing tours or packages, suggesting AI could influence how travellers discover and assemble travel experiences.

The study also highlighted the potential risks of advertising within AI-generated answers. Most participants encountered no sponsored content during the research period, but irrelevant advertisements were found to undermine the credibility of the wider AI response.

The findings point to an opportunity for travel suppliers to attract customers directly through AI recommendations, potentially reducing their reliance on traditional intermediaries.

However, the transition from AI-powered research to booking remains a challenge.

Although participants were generally willing to act on AI recommendations, many preferred to retain control of the payment process. The majority of booking handoffs failed to preserve essential details, including travel dates and guest numbers.

Participants tended to attribute problems with the booking process to the travel company rather than the AI tool, suggesting that poor integration between AI platforms and booking systems could affect the customer experience without necessarily undermining trust in AI itself.

The findings reinforce the importance of ensuring that travel companies can convert AI-driven discovery into a straightforward booking experience.

Propellic said hotels, destinations, tour operators and OTAs should prioritise their visibility across AI platforms by regularly testing how their brands are represented in ChatGPT, Gemini and Google AI Mode.

The agency recommends structuring website content so AI tools can easily extract and compare information, providing clear pricing, terms and business details, and building review volumes and photographic content on platforms referenced by AI.

It also advises travel companies to test the transition from AI recommendations to their booking engines, ensuring customers can easily resume the booking process if information is lost.

The report further suggests that travel brands focus on relevance rather than reach when considering advertising opportunities within AI-generated answers.

The findings come as AI becomes an increasingly prominent part of the travel planning process. 

Propellic cites Phocuswright research indicating that AI featured in 56% of travel purchase journeys in the first half of 2026, up from 33% a year earlier.