ChatGPT Ads Won’t Kill OTAs. But They Could Kill Direct’s Advantage

When OpenAI confirmed that ads are coming to ChatGPT, much of the travel industry focused on placement, labeling, and philosophical questions about AI influence. That conversation misses the point.

The more important shift is structural. ChatGPT ads introduce a new demand channel built on conversational intent, not search queries. And the brands that learn how to operate inside that environment first will define how value is captured inside it.

If history is any guide, Online Travel Agencies will move faster than most travel brands.

Direct booking has a real edge today

Early research in partnership with Phocuswright shows that nearly 80% of bookings driven by LLMs (Google AI Mode in this specific study) are currently happening direct, not through OTAs. That is an unusually strong signal in an industry where intermediaries typically dominate new performance channels over time.

This is happening because large language models tend to reward clarity and confidence over coverage. When a user asks for a recommendation, the model often surfaces brands it can explain cleanly - clear value propositions, recognizable names, straightforward policies, and simple booking paths.

In other words, LLMs currently favor brands that are easy to describe and easy to trust.

For hotels and travel brands, this has quietly created a window of advantage. Direct bookings benefit from brand familiarity, reduced friction, and the perception of authority when the answer itself feels personalized.

But this advantage is not durable on its own.

Conversational ads change the economics

Ads inside ChatGPT do not behave like display or social ads. They will behave more like search ads, but they sit adjacent to a moment of decision, inside a guided experience where the user has already expressed intent. 

That environment compresses the funnel. There is less browsing, less comparison, and less patience for ambiguity. Performance matters immediately.

This is where OTAs historically thrive. They are built to:

• Translate intent into inventory instantly
• Test formats at scale with low marginal cost
• Optimize toward conversion rather than storytelling
• Accept short-term inefficiency in exchange for long-term learnings

Conversational AI doesn’t remove those advantages. It amplifies them.

Muscle memory is the real moat

OTAs did not win Google because they had better ads. They won because they built institutional reflexes early.

For example, Booking.com developed a sophisticated in-house experimentation platform that enabled thousands of concurrent A/B and multivariate tests across marketing and product. That capability allowed them to adapt faster than brands relying on slower decision cycles and external vendors.

Over time, OTAs learned how to:

• Interpret new signals before they were well defined
• Adjust bidding, creative, and landing paths weekly
• Feed platforms the data those systems implicitly rewarded
• Scale what worked faster than individual brands could react

That muscle memory compounds. Once a channel matures, late entrants are no longer learning the rules. They are playing inside them.

ChatGPT ads represent the beginning of that same cycle.

The channel is immature - but that favors early movers

It is still unclear exactly how ads will be placed, priced, or measured inside ChatGPT. Early indications suggest OpenAI is piloting with a limited set of partners rather than opening a fully self-serve marketplace on day one.

That uncertainty may cause some brands to hesitate. It shouldn’t.

Historically, the period of ambiguity is when the most durable advantages are built. Early partners learn how prompts map to outcomes, what types of inventory convert, and how conversational framing changes user behavior. By the time the rules are fully published, the playbook already exists.

OTAs are structurally incentivized to participate early, even when the economics are imperfect. Most travel brands are not.

AI tooling should help brands - but only if they use it

In theory, AI-driven product development should lower the barrier for brands to compete. Better tooling, faster experimentation, and more accessible insights should reduce the structural advantage intermediaries have traditionally enjoyed.

But tools do not create muscle memory. Practice does.

Brands that assume AI products alone will level the playing field without early operational investment are repeating a familiar mistake. The advantage does not come from the software. It comes from learning how to deploy it under real demand conditions.

Measurement will lag, but learning won’t

One of the reasons brands hesitate to engage early is uncertainty around attribution. That discomfort is understandable, but it is also dangerous.

OTAs are comfortable operating with imperfect attribution. In practice, they commit to a directional source of truth so decisions can be made quickly. They optimize on cohort behavior, assisted conversion, and long-term lift, while making investments into more sophisticated tracking tools like Media Mix Modeling (MMM) to understand incrementally caused by a new channel rather than directly attributed sales.

Many brands wait for clean dashboards before committing. In conversational AI, that hesitation is a competitive disadvantage.

The call to action for travel brands

This is not a moment for passive observation. It is a narrow window to build operational familiarity before the channel calcifies.

Travel brands should be doing four things now:

• Building muscle in adjacent paid channels while access is limited
  If ChatGPT ads are not yet available, brands should be pressure-testing high-intent paid media now - search, metasearch, and upper-funnel formats where intent collapses quickly. The goal is not efficiency. It is learning how fast experimentation, prompt-adjacent messaging, and rapid iteration work in practice before LLM ads fully open up.
  
  
• Doubling down on SEO and GEO to defend organic share
  Brands should assume that organic visibility inside LLMs will remain meaningful, but increasingly competitive. That means investing in structured content, machine-readable differentiation, and authoritative signals that help models explain why a brand is the right answer - not just that it exists.
  
  
• Strengthening AI-readable signals like policies, value props, and differentiation
  LLMs reward clarity. Cancellation policies, inclusions, loyalty benefits, and experiential differences need to be explicit, consistent, and easy to summarize conversationally.
  
  
• Reducing booking friction aggressively
  Conversational demand punishes hesitation. Fewer steps, fewer surprises, faster confirmation, and clearer pricing are no longer UX optimizations - they are conversion requirements.
  
  

Because if brands do not build muscle memory around LLM-driven demand, OTAs will.

And once intermediaries define how conversational ads work at scale, direct booking’s current advantage will not disappear slowly. It will disappear all at once.

History has already shown how this story ends. The only open question is whether travel brands choose to act before it repeats.

‍