Booking Holdings Says AI Cut Customer Service Costs 10%—Investors Cheer, Staff Brace

The number appears clean on paper. Using generative AI, a 10% reduction in customer service costs per reservation was demonstrated during an earnings call with the composed assurance of a business aware that investors were paying close attention. It sounds like progress in the low-key language of corporate performance. But what they rearranged to exist is frequently left out of numbers, particularly round ones.

The mood seems more difficult to measure inside Booking Holdings, where glowing monitors bounce off glass partitions, and travel posters still feature beaches that the majority of employees hardly ever visit. Previously crowded with overlapping voices in Dutch, English, Spanish, and occasionally irritated German, customer service floors now feel more organized. Maybe too orderly. One gets the impression that something has been overly simplified.

The leadership of the company presented the change as a technological advancement, citing increased bookings and growing profit margins. According to executives, there is still a high demand for travel, and AI is enabling the business to grow without allowing expenses to skyrocket in tandem with expansion. This appears to be the next logical step for investors. After all, automation has already changed a number of industries, including hotel check-ins and airline ticketing counters.

However, those who are closest to automation rarely perceive it as neutral.

Some of the former support staff still congregate at cafés close to the old office buildings in Amsterdam, where Booking.com evolved from a tenacious European startup into a global travel gatekeeper. They frequently return to the same point in their conversations, which veer between nostalgia and anxiety: the first time they noticed that customers were talking to software more often than to them.

It wasn’t abrupt. Seldom is it.

Initially, AI answered basic queries like confirmations, date changes, and refund requests. Time-consuming repetitive tasks that served as the foundation for entire careers. Slowly, the equilibrium changed. Less frequently, human agents intervened, only being called upon when discussions became emotional or complex. It seems as though the work itself became more focused on catching what machines couldn’t do and less about assisting people.

Whether this makes the work better or just less common is still up for debate.

These efficiencies were crucial to the company’s transformation program, which resulted in savings totaling hundreds of millions. Executives discussed reinvesting in platform enhancements, loyalty programs, and artificial intelligence. It was difficult to ignore how well the story flowed while watching the earnings call. Innovation is financed by efficiency. Innovation propels expansion. Shareholders benefit from growth.

Silence was something that the story didn’t dwell on.

In the past, customer service was intentionally noisy. Voices overlapping, phones ringing, and tension fluctuating throughout the day. Nowadays, a large portion of that friction occurs during unseen interactions between algorithms and users. Customers frequently hear a serene digital voice offering options with courteous precision when they first call.

Some clients are unaware of the distinction.

Others do.

Before speaking with a human, a travel writer in Barcelona explained that she had to spend almost an hour attempting to explain a fraud issue to an AI agent. The AI wasn’t impolite. It wasn’t irritable. It was just incapable of comprehending urgency. The exchange resembled filling out an endless form more than having a conversation.

Efficiency felt strangely unconcerned at the time.

In this change, Booking Holdings is by no means alone. Businesses in the tech and service sectors are learning that AI can perform tasks that were previously thought to require human judgment. Startups guarantee that their customer service representatives will perform better than entire teams. Executives freely discuss how AI works like “ten times employees.”

At least mathematically, they might be correct.

Customer service, however, was never just a math problem.

There’s a reason why people used to call people by name when they were stuck in strange cities or had reservations that had been stolen. They weren’t looking for quicker responses. They wanted to feel reassured. While software can offer solutions, automating reassurance appears to be more difficult.

As they observe Booking’s growing profits and margins, investors seem content for the time being. The business produced billions in free cash flow and gave shareholders sizable returns. The financial reasoning is still convincing. reduced expenses. Larger margins. operations that are more predictable.

Nevertheless, as this develops, it seems like a cultural shift is taking place within organizations like Booking.

In the past, customer service served as a conduit between businesses and the erratic emotional realm of their clientele. On the other hand, AI transforms emotion into structured inputs. Issues turn into information. Discussions turn into transactions.

It works well. And perhaps the objective was always efficiency.

However, one unused headset remains hanging next to a desk in the corner of a former Amsterdam customer service office. No one bothered to take it off. Dust accumulates gradually, as it does during silent transitions.

The headset is still in use today. It is merely awaiting a type of work that may not come back.

There is proof to back up Booking Holdings’ assertion that AI is improving service while cutting costs. Reservations are now open. The margins are getting better. Every day, millions of reservations are still made by customers.

However, beneath the neat geometry of corporate metrics, a different, more difficult-to-chart reality is emerging.

Efficiency makes a noise. It also sounds like there are fewer voices at times.