Case Study: Voxpopme achieves 1,000% performance boost with Elastic

Leveraging Elasticsearch for a 1,000% performance boost

Voxpopme is a video-insight platform that ingests consumer-recorded video and long-format content to deliver searchable sentiment, themes and showreels to brands. Rapid growth (half a million surveys in 2017) exposed limits in a legacy monolithic PHP stack that split data across MySQL, MongoDB and a small Elasticsearch index. Complex user queries required multiple sequential database calls and ID intersections, turning sub-second lookups into operations that could take up to 30 seconds and hurting scalability.

The team rewrote search to flatten and index each record as a single Elasticsearch document on every write, letting any user query be handled by one ES request and matching the API’s JSON output. After moving from AWS Elasticsearch to Elastic Cloud (with X‑Pack) and optimizing indexing (bulk writes, field reallocation), they reported ~1,000% performance gains, query latencies dropping from seconds to as low as ~2 ms, memory pressure falling from ~75% to ~25%, plus new logging and aggregation features for users — and plans to expand Elasticsearch for internal analytics.

Voxpopme

David Maidment

Senior Software Engineer