Why Google Shops and Amex GBT Books: Revenue per Employee Explains the Travel Industry's Balance

Sabre’s CEO said this month that Google could get full booking capability “en masse tomorrow.” And Google, as it has for years, says it does not want to be an OTA.

I heard that explained as strategy for a long time. Google is careful. Google does not want the liability. Google does not want to run a call center. All of it plausible, none of it the reason.

This week I pulled four 2025 annual reports and did one calculation. Revenue divided by employees.

The ladder

	Employees (Dec 31, 2025)	Revenue per employee
Alphabet	190,820	$2.1M
Booking Holdings	24,300	$1.1M
Expedia Group	16,000	$0.9M
Amex GBT	over 27,000	about $100K

Source: FY2025 Form 10-K filings for Alphabet, Booking Holdings, Expedia Group and Global Business Travel Group.

Read it as a ladder, not a list.

Google sits at the top and shops. Search, compare, rank. That scales with compute, and Google has more compute than anyone. Almost no people per booking, because there is no booking.

Booking and Expedia own the middle rung: hotel fulfilment. A hotel booking is confirm, cancel, done. It needs people, but not many, and Booking’s 10-K shows exactly where it is willing to go next: flight gross bookings grew 29% in 2025. Leisure air, where fulfilment is still mostly self-serve. Not corporate.

Amex GBT owns the bottom rung. Corporate flights, and everything that happens after the ticket is issued. The change. The cancellation at 11pm. The traveler stuck in the wrong city with a policy that says no and a manager who says yes. Somebody has to pick up the phone, and that somebody is one of 27,000 people.

Amex GBT has more employees than Booking Holdings and makes about twenty times less revenue per person than Alphabet. Not because it is a worse company. Because of where it sits on the ladder.

Nobody is being polite

The usual story is that these companies stay in their lanes out of strategy or caution. The numbers say something simpler. Every rung down the ladder, fulfilment needs more people per dollar of revenue, and each player stopped exactly where its own math stopped.

Google is not going to swallow a business that runs at $100K per head. It would take a company that earns $2.1M per employee and bolt on a division that earns twenty times less. So it shops, hands the booking to Amadeus and Sabre and the OTAs, and calls it partnership.

Booking is not going into managed corporate travel either. Its per-head economics are built for confirm-cancel-done, and every step toward servicing pulls the ratio down.

And Amex GBT does not need to worry about either of them coming down the ladder, because the thing that protects it is the thing that makes it expensive: 27,000 people.

This balance has held the entire time I have worked in this industry. I sat in rooms where it was treated like the weather. Something that simply is.

The one number that would change it

Here is what most of the AI-in-travel conversation gets wrong. It talks about shopping. Better search, better recommendations, an agent that finds you the flight. AI already owns that layer, and the ladder did not move, because shopping was never where the balance lived.

The only variable that matters is servicing headcount per booking. How many people does it take to stand behind a corporate itinerary from the moment it is ticketed until the traveler is home.

I run AI agents on production Sabre inside real travel agencies, and I will be honest: I do not have that answer yet. The demos say yes. The agent rebooks the missed connection, re-queues the PNR, tells the traveler before they know there is a problem. The 11pm cancellation, at volume, with a real client on the other end, says come back in a year and show me the numbers.

If it moves

But suppose it moves. Not in 2030. In the next twelve months.

Then the balance does not shift in favor of one player. It disappears for all of them on the same day, because the reason every player stopped where it stopped is the same reason.

Google can step down a rung, because booking no longer drags its ratio. Booking can walk into corporate air. And a company with zero legacy headcount can walk straight onto the bottom rung with a cost base that 27,000 people cannot answer, no matter how good those people are.

I do not know who that will be. I know the ratio is the thing to watch, not the announcements. When someone publishes servicing headcount per booking and the number has moved, that is the day the industry changes.

Which rung do you think moves first?

This is the number we watch every day while building Bitravel. Alex, our agent, runs on production Sabre inside real agencies, and the honest status is that we are still measuring. If you run an agency and want to see what that looks like on a live PNR, book a 30-minute call. No deck.