Using AWS helped us to simplify our architecture, reduce code complexity, and most importantly develop features quickly to improve user engagement.
Prakash Janakiraman Cofounder and Chief Architect

Social-network innovator Nextdoor is running its big-data and analytics solution on AWS, resulting in a 20 percent conversion on notifications—a key indicator of customer use—along with a 15 percent increase in unique content viewed per member.

The free and popular service is available in more than 115,000 neighborhoods across the country and growing every day. But that growth posed challenges for the company, so it turned to AWS.

“Our members and city agencies are increasingly using Nextdoor for time-critical updates for disaster recovery, crime and safety, local events, and more,” says Prakash Janakiraman, Nextdoor’s cofounder and chief architect. “Managing the increasing volume of content and email-push notifications became unwieldy for members. We wanted to improve the user experience by routing only the ‘best’ content notifications more effectively, which required us to examine and respond to behavioral data in real time. It had to be easy to operate and to understand, extend, and scale.”

Nextdoor used machine learning models, Amazon EC2 Container Service (Amazon ECS), Amazon Elastic MapReduce, AWS Lambda, Amazon Kinesis, and Spark to create a “trending post” feature that notifies users of the top daily discussion in their neighborhood. Data is analyzed, transformed, and deposited into an Amazon DynamoDB data store that can be accessed by live production systems. Notification systems then read aggregate data and determine who should be notified. Additionally, the company found ways to improve its newsfeed experience by prioritizing more “interesting” conversations above others.

See more customer stories