Booking.com drives 65% increased AI adoption with DX

Booking.com, one of the world’s largest online travel platforms, serves more than 500 million monthly visitors and processes over one billion bookings each year. As part of a broader effort to drive efficiency across the company, Booking began rolling out an AI code assistant—but adoption fell short of expectations.

“We introduced an AI code assistant from a vendor we already worked with, so we thought it would be straightforward for developers to adopt,” explains Bruno Passos, Group Product Manager for GenAI and Developer Experience. “But developers were hesitant. We started to hear complaints about the tool and saw signs that usage wasn’t where we thought it would be.”

To make things worse, Booking lacked visibility into how developers were engaging with the tool or why they weren’t. They wanted deeper insights into adoption and utilization, specific types of tasks AI was being used for, and impact on productivity. “It’s critical to collect both quantitative and qualitative data when rolling out a tool like this,” says Zane Wright, Senior Product Manager. “You need data to understand where tools are helping or hurting, and how developers feel about using them.”

Booking had already been using DX to measure overall developer productivity. To get specific insights on AI usage, Booking expanded their dataset in DX to include metadata from their AI code assistant, which enabled rich analysis and correlation to key metrics around velocity, quality, and developer experience. “Once we brought in DX, we were finally able to get answers,” says Wright. “It gave us a direct line of sight into usage and sentiment that we didn’t have before.”

DX data revealed that daily active users of their AI tool had 16% higher change throughput than non-users. It also identified hotspots where training and enablement were lacking amongst non-users. “We realized that enablement was a missing element,” says Passos. “People needed support to understand how to use the tool effectively.”

Armed with data from DX, Passos’ team launched several targeted initiatives, including a series of two-day workshops that paired foundational GenAI training with real-world problem-solving, regular office hours for hands-on support, and ongoing development of content to highlight new AI capabilities and best practices.

Thanks to these efforts, today, nearly all of Booking’s engineers are using AI tools. Booking has also seen an additional 15% increase in change throughput (31% improvement from the original baseline). “People are like, ‘We love it’ now,” says Passos. “They want to try every AI tool under the sun.”

Now that Booking has achieved widespread adoption, they’re focusing on driving more regular usage. “We’ve seen in DX that developers who use AI on twelve or more days per month are significantly more effective. So we have shifted our goal from adoption to daily adoption,” says Passos. “That’s what’s next for us.”

“DX really helped us take our GenAI journey further,” says Wright. “It showed us where to focus and helped us get way more impact out of the tools, both in how deeply and widely they’re being used.”