Case Study: Düğün.com improves wedding search and discovery with Amazon Web Services

Dugun.com cuts p99 search latency 51% with Amazon Web Services

Düğün.com, a large wedding marketplace headquartered in Türkiye, faced challenges with its outdated machine learning infrastructure. Its search function could not understand the intent behind couples' natural language descriptions, causing relevant vendors to be missed. The company's hand-deployed ML scripts also required rebuilding foundational elements for every new project, hindering development.

Amazon Web Services provided the solution by enabling Düğün.com to rebuild its data science layer as shared services. Using Amazon Bedrock, the company developed a natural language search feature and an image-to-video conversion tool for gowns. This resulted in a 51% reduction in p99 search latency, a 170ms average warm response time, and a doubling of conversions and click-through rates with video clips, all while personalizing the wedding planning experience.

Dugun.com