Let's Meet Locally powers location-aware ecommerce experiences for over 47,000 retailers worldwide

Company

Locally is a leading shopping enablement platform that connects brands, retailers, and consumers. Founded in New Orleans in 2014, Locally serves over 55 million unique online shoppers each month and facilitates annual online-to-offline sales referrals worth over $1.8 billion. The platform represents more than 47,000 retailers worldwide and sits directly in the critical path of many high-traffic brand and retailer websites, handling highly dynamic and strongly seasonal traffic patterns that place strict requirements on performance, reliability, and cost control. Locally enables shoppers to find and reserve products they research online for pickup at local stores, bridging the gap between online research and in-store shopping.

Challenge

Locally had been running Elasticsearch since its early days in 2014. Over time, the team found that the operational overhead, cost, and indexing performance at scale made it harder to evolve the platform efficiently. The engineers found Elasticsearch unapproachable, which prevented sufficient momentum in development. Additionally, Elasticsearch was incredibly complicated and expensive to host and maintain, and indexing data at their scale was too slow.

Ben Hirsch, CTO & Co-founder: "We began tinkering with Manticore in early 2024 and immediately realized how fast it was — comparable to Elastic at runtime and much faster indexing. We then deployed a proof-of-concept to production and noticed a 16× reduction in cost from equivalent Elasticsearch services."

Solutions and results

• Locally migrated from Elasticsearch to Manticore Search, achieving a 16× cost reduction while maintaining comparable query performance.
• Manticore Search now processes tens of thousands of queries per second in production across multiple workloads: aggregations, geo-spatial indexing, ecommerce search, and vector-powered typeahead.
• The deployment on Google Cloud Platform uses Compute Engine instances with auto-scaling, managed instance groups, and HAProxy load balancers for optimal performance and reliability.
• The system has successfully handled two holiday seasons with minimal issues and virtually no downtime, demonstrating stability under peak seasonal demand.
• Day-to-day operations are now fully automated and hands-off, allowing the team to focus on product development rather than maintaining search infrastructure.

Read more in our blog .