Glad you enjoyed the thought exercise! Deep in the weeds on evaluating different actors across the stack and will be posting a more detailed mapping including where we think the biggest potential for rate-of change is (both positive and negative)
1. The whole model hinges on agents charging BKNG almost nothing (under 1.25%) instead of Google's 27%. But there's yet no reason to believe that. Every gatekeeper in tech history (i.e. Apple's App Store, Google itself) ends up taking a big cut once it controls the customer relationship. If 2-3 AI agent platforms end up owning "how people book travel," they'll probably act exactly like Google did.
2. Google does not want paid search to die, so I'd assume that if people just ask Google's own AI to book their trip instead of googling it, they will change that to stop running an ad auction and start running a toll booth instead.
3. Agents might not be neutral. The model assumes agents just pick the best deal for the user. In reality, whoever builds the agent (OpenAI, Google, Apple) will have their own business incentives (i.e. placement deals, preferred partners) just like "unbiased" price comparison sites have never really been unbiased.
4. Loyalty programs cut against BKNG. If your AI agent is actually working for you, it might book directly with Marriott or Hilton to keep your points and status, skipping BKNG entirely.
5. BKNG's own "we handle trust and refunds" moat. If trust becomes a built-in feature of the payment rail itself, BKNG's "we're the trusted middleman" advantage weakens too.
Overall, really good mental exercise, and I'm just trying to play devils advocate as I also think about the stock and the overall idea :)
What other aggregator/platforms caught your attention? Great post
Glad you enjoyed the thought exercise! Deep in the weeds on evaluating different actors across the stack and will be posting a more detailed mapping including where we think the biggest potential for rate-of change is (both positive and negative)
Great post! Some thoughts:
1. The whole model hinges on agents charging BKNG almost nothing (under 1.25%) instead of Google's 27%. But there's yet no reason to believe that. Every gatekeeper in tech history (i.e. Apple's App Store, Google itself) ends up taking a big cut once it controls the customer relationship. If 2-3 AI agent platforms end up owning "how people book travel," they'll probably act exactly like Google did.
2. Google does not want paid search to die, so I'd assume that if people just ask Google's own AI to book their trip instead of googling it, they will change that to stop running an ad auction and start running a toll booth instead.
3. Agents might not be neutral. The model assumes agents just pick the best deal for the user. In reality, whoever builds the agent (OpenAI, Google, Apple) will have their own business incentives (i.e. placement deals, preferred partners) just like "unbiased" price comparison sites have never really been unbiased.
4. Loyalty programs cut against BKNG. If your AI agent is actually working for you, it might book directly with Marriott or Hilton to keep your points and status, skipping BKNG entirely.
5. BKNG's own "we handle trust and refunds" moat. If trust becomes a built-in feature of the payment rail itself, BKNG's "we're the trusted middleman" advantage weakens too.
Overall, really good mental exercise, and I'm just trying to play devils advocate as I also think about the stock and the overall idea :)
Large chains disintermediation booking in an agentic scenario is a counter and googLe’s built in agentic search
Yes agree we highlight the large chains as MoR competitors and where BKNG wins on aggregating the tail + query layer (not so diff from today)
Google building agentic search just puts them in competition with all the other AI platforms, it does not substitute for their current monopoly