Case Study: Workday achieves dramatic improvements in search relevance, precision, and scalability with Elastic

Workday - Customer Case Study

Workday, the enterprise cloud provider for finance and HR, faced a search-relevance crisis in its recruiting product: users were getting thousands of noisy hits from unstructured resumes and cover letters, relevance metrics were low, and the search infrastructure was becoming unmanageable and costly to scale (large cluster state, many tenant indices and security/upgrade concerns). The Search & ML team undertook user research to understand real queries and use cases and defined the core problem as finding precise, relevant candidates rather than just broad matches.

The team chose Elasticsearch, implemented document parsing to extract structured fields (name, title, skills), tuned ranking for deep vs. shallow search needs, consolidated tenant indices to cut cluster-state bloat, and published tooling (e.g., elasticrypt) to address security and operational unknowns. Results were dramatic: total hits for target queries fell from ~10,000 to ~100, precision@50 rose from 0.2 to 1.0, NDCG for name/job title/skill improved roughly from 0.1 to 0.9/0.85/0.65, and cluster state shrank from ~800 MB to ~20 MB, enabling much better relevance, scale, and maintainability.

Workday

Angela Juang

Senior Engineer