Why OTAs Are Investing in AI and What It Means for Direct Bookings

The OTAs are not waiting

Booking.com has its AI trip planner. Expedia has integrated ChatGPT into its platform. TripAdvisor is building conversational search. Google is weaving AI Overviews with booking links into the search results that billions of travellers use every day.

For independent hospitality businesses that have spent years building direct booking capabilities, this is the next front in a long-running battle. And unlike the SEO and paid search competition that defined the last decade, this one is harder to see, harder to measure, and potentially harder to win.

At Delphium Labs, we have been tracking how OTA AI investments are reshaping discovery and booking patterns. The data suggests that the direct booking gains many independents achieved through better websites and SEO are at risk unless those businesses adapt to a fundamentally different discovery model.

What OTAs are building

The major OTAs have committed significant resources to AI-powered features. Understanding what each is building is essential context for any direct booking strategy.

Booking.com AI Trip Planner

Booking.com launched its AI trip planner in 2024 and has expanded it aggressively through 2025 and into 2026. The tool allows users to describe their ideal trip in natural language and receive a complete, bookable itinerary. A user can say "I want a four-night break in the Scottish Highlands, staying somewhere cosy with a good restaurant, for under 150 per night" and receive a curated set of Booking.com properties with direct booking links.

The critical point: the AI trip planner only recommends properties listed on Booking.com. It draws from Booking.com's own database of reviews, ratings, and property descriptions. If your property is listed on Booking.com, it may appear. If it is not, it will not. Either way, any booking through this tool generates OTA commission.

Expedia ChatGPT integration

Expedia was among the first OTAs to integrate directly with ChatGPT, initially through a plugin and now through a deeper partnership that surfaces Expedia-sourced recommendations within ChatGPT conversations. When a ChatGPT user asks about travel options and Expedia data is relevant, the AI engine can present Expedia-sourced results with booking links.

Delphium Labs testing in early 2026 found that approximately 15% of hotel-related ChatGPT responses included Expedia-sourced information, up from roughly 8% six months earlier. The integration is becoming more prominent, not less.

TripAdvisor conversational search

TripAdvisor has rebuilt much of its search experience around conversational AI. Users can ask questions in natural language and receive responses that draw from TripAdvisor's massive review database. The tool synthesises thousands of reviews into concise recommendations, making TripAdvisor's review data even more central to how travellers choose properties.

For businesses with strong TripAdvisor profiles, this can be positive. For those without, it is another discovery channel that favours established, well-reviewed properties.

Google AI Overviews with booking links

Perhaps the most significant development is Google's integration of AI Overviews with hotel and restaurant booking capabilities. When a user searches for accommodation or dining, Google's AI increasingly generates a summary answer that includes specific property recommendations, pricing, and direct booking or reservation links. These AI Overviews sit above traditional search results, capturing attention and clicks before users ever scroll to organic listings.

Delphium Labs monitoring shows that AI Overviews now appear for an estimated 40% of UK hospitality search queries, and that percentage is growing month over month. For many of these queries, the AI Overview includes links to Google Hotels or Google Maps, which typically feature OTA pricing alongside direct rates.

Why this matters for independents

In the traditional search era, independent properties could compete with OTAs through SEO and targeted Google Ads. A well-optimised website could rank alongside or above OTA listings for key search terms. The direct booking movement built on this foundation: invest in your website, optimise for search, and capture bookings without paying commission.

AI-powered discovery changes the equation in three important ways.

The OTA's answer becomes the recommendation. When a traveller uses Booking.com's AI trip planner, the recommendation is inherently an OTA-mediated one. There is no opportunity for a direct booking in that interaction. The same applies when ChatGPT surfaces Expedia-sourced data or when a Google AI Overview links primarily to Google Hotels. The recommendation layer sits between the traveller and the property, and the OTAs control it.

Organic discovery carries commission risk. This is the subtlety that many operators miss. If a traveller asks ChatGPT for hotel recommendations and receives an answer sourced from Expedia data, the booking path runs through Expedia. The traveller never visited a search engine, never saw your website, and never had the option to book direct. What looks like organic AI discovery is, in practice, OTA-mediated booking with full commission attached.

SEO alone cannot solve this. Traditional search optimisation improves your ranking in a list of results. AI-powered discovery does not always present a list. It presents a synthesised answer. Being well-optimised for traditional search may not translate into visibility within AI-generated responses, especially when those responses draw from OTA data rather than your own website.

The direct booking angle

The picture is not entirely bleak. Delphium Labs testing shows that properties with strong direct content can still win in AI engines, even when OTAs are competing for the same queries.

Our research found that properties with rich, detailed direct booking pages get cited alongside, and sometimes instead of, OTA listings. The key finding: when a property's own website contained more detailed, more specific content than the corresponding OTA listing, AI engines were 1.7 times more likely to cite the direct source.

This makes sense when you understand how AI engines work. They are looking for the most relevant, most detailed information to answer a user's query. If your website has a 300-word room description with specific details about the view, the furnishings, and the local area, while your Booking.com listing has a generic 50-word summary, the AI engine has more useful material from your direct site.

The differentiator is unique content that OTAs cannot replicate. OTA listings are constrained by standardised templates. Your own website is not. The content that wins in AI discovery is precisely the content that OTAs cannot match: local expertise, personal stories, detailed descriptions, and contextual information about the surrounding area.

What to do about it

Based on Delphium Labs research and ongoing monitoring, here are the specific actions that give independent properties the strongest position in AI-powered discovery.

Build content that OTAs cannot match

The most effective defence against OTA dominance in AI discovery is content differentiation. OTA listings follow templates. They contain standardised descriptions, professional but generic photos, and review aggregations. Your website can offer what they cannot:

Local area guides. Write detailed, opinionated guides to your area. Best restaurants within walking distance. Hidden attractions that guests love. Seasonal recommendations. This is exactly the type of content that AI engines surface when travellers ask contextual questions like "hotel near the best food scene in Bristol".

Personal and specific stories. The history of your building. How you source your ingredients. Why you chose that particular artwork. AI engines increasingly value unique, first-person content because it is harder to find elsewhere and more useful for answering specific queries.

Detailed room and facility descriptions. Our own testing at Delphium Labs confirms that specificity wins. Room dimensions, view descriptions, bed types, bathroom details, in-room amenities listed individually. AI engines draw from this material when generating recommendations, and the more detail you provide, the more likely you are to be cited.

Implement schema markup that points to your site

Schema markup tells AI engines what your content means, not just what it says. For hospitality businesses, implementing comprehensive schema is one of the highest-impact technical actions available.

Crucially, your schema should include your direct booking URL. When AI engines process structured data that includes a booking link pointing to your own website, that link has a stronger chance of appearing in the AI response. If your schema only references OTA booking links, you are directing AI-mediated traffic away from direct bookings.

Delphium Labs recommends implementing LodgingBusiness or Restaurant schema with complete property details, AggregateRating with review counts, room or menu schema with pricing, and direct booking URLs prominently included.

Optimise for queries where OTAs are weakest

OTAs dominate generic queries like "hotels in London" or "best hotel deals Edinburgh". They are far weaker on specific, contextual queries:

• "Dog-friendly pub with rooms near the Peak District with good walking routes"
• "Wedding venue in the Cotswolds that does its own catering"
• "Boutique hotel in Bath with a good spa and restaurant for a 30th birthday"

These long-tail, intent-rich queries are where independent properties can compete effectively. Delphium Labs analysis found that OTA dominance drops significantly for queries that include three or more specific criteria. For these queries, properties with rich, relevant website content often outperform OTA listings in AI responses.

Building content that directly addresses these specific queries is one of the most effective strategies available.

State direct booking incentives clearly

If your website offers a best-rate guarantee, complimentary extras for direct bookers, or any other incentive, state it prominently and clearly. AI engines can and do pick up on direct booking incentives mentioned on property websites. In Delphium Labs testing, properties that explicitly stated "book direct for our best rate" or "direct booking includes complimentary breakfast" on their websites were more likely to have those incentives mentioned in AI-generated recommendations.

Make these incentives visible in your content, not buried in terms and conditions. Write them as clear statements that an AI engine can extract and present to a user.

The bigger picture

The OTA investment in AI is substantial and accelerating. Booking Holdings allocated over $1 billion to AI development in 2025. Expedia Group has made AI its strategic priority. Google continues to expand AI features across search and maps.

Independent properties cannot match this investment. But they do not need to. The advantage independents hold is authenticity, specificity, and unique content that no OTA can replicate at scale. The businesses that lean into this advantage, that build websites rich with genuine, detailed, locally-rooted content and support it with proper technical infrastructure, are the ones that will maintain and grow their direct booking share.

AI discovery could either deepen OTA dependency or be the tool that breaks it. The outcome depends on what independent properties do now. The data from Delphium Labs research is clear: those that invest in content quality, structured data, and AI-specific visibility strategies are already seeing results. Those that rely solely on traditional SEO and hope that AI will work the same way risk losing ground that is very difficult to recover.

FindingFin helps identify which queries still have a path to direct discovery and where OTA-sourced answers are dominating. Understanding your current position is the first step. The businesses that measure, adapt, and invest in the right areas will be the ones that protect their direct booking revenue in the age of AI-powered travel discovery.