Generating Dynamic Ads in Under 1 Second Using AWS Lambda with Booking.com

Booking.com’s ad-tech team manages the company’s social media advertising campaigns, creating dynamic ads that showcase properties for potential guests. To create engaging ads, the team needs to automatically add dynamic content like prices, location details, and branded elements on top of property images. The team was using a third-party vendor for that work, but the arrangement was costly and limiting.

“We started asking ourselves: Do we want to continue using the third-party vendor, or would we get more visibility if we owned the whole solution from start to finish?” says Niharika Gupta, senior software engineer at Booking.com. “We also wanted to reduce costs, because using a third-party company for creating these overlays meant paying them millions of dollars to do this work for us.”

The third-party solution restricted how the team could design and customize ad templates, limiting its ability to test new creative approaches or quickly adapt to changing market needs. Additionally, the complex integration between systems made it difficult to monitor performance or quickly address issues when they arose. After evaluating potential alternatives, the ad-tech team realized that a serverless architecture could provide the scalability and flexibility that it was looking for while also reducing costs. Booking.com’s leadership had identified AWS as a strategic provider for its broader cloud transformation, so the ad-tech team looked to AWS services as the foundation of the new solution.

“Using AWS, we could build the whole solution as serverless,” says Gupta. “This was actually the first time my team and I thought we could build something at such a big scale using entirely serverless solutions, and all the services that AWS provides out of the box, without worrying about small configurations. That was the major reason we wanted to shift toward serverless services on AWS.”

Booking.com set out to build a scalable image rendering service that could support thousands of requests per second with low latency and high availability. Booking.com has over 1.5 million hotels around the world available for booking, so the solution also needed to support content in multiple languages and currencies. The ad-tech team also wanted to support marketing specialists in quickly testing and optimizing creative elements of specific ads in near real time. After engaging the AWS team for architectural guidance, Booking.com developed and put its in-house solution into production in early 2024.

When a social media channel needs an image for an ad, it sends a request to Amazon CloudFront, a service that securely delivers content with low latency and high transfer speeds. An API Gateway forwards the request through Booking.com’s security layer, preventing unauthorized external access.

Then, the request activates AWS Lambda, a serverless compute service that runs code in response to events and automatically manages the compute resources. The AWS Lambda functions then retrieve the base hotel image along with the required overlay elements—such as logos, fonts, and templates—from Amazon Simple Storage Service (Amazon S3), an object storage service built to store object data everywhere. Using these components, the solution uses AWS Lambda to create the final image. The dynamic elements, such as current pricing and location information, are layered on top of the property photo. The completed image is then delivered back through Amazon CloudFront to the social media partner.

The ad-tech team relies on Amazon CloudWatch, a service that observes and monitors resources and applications on AWS, to gain complete visibility into system performance. “Before, we had zero visibility,” says Gupta. “Using a managed solution on AWS, we went from zero to 100. And if things go wrong we can quickly intervene to fix it.”

Each feed contains 1.5 million hotel listings with 10 images each, and Booking.com manages 200 different feeds for up to 10 social media partners. Despite this volume, the solution maintains impressive performance: It processes over 1,000 requests per second and delivers images in less than 1 second on average, with 99.9 percent availability. Because Booking.com can automatically scale based on demand, the company pays only for what it uses. That efficiency, combined with the removal of third-party fees, reduced costs by 90 percent.