Case Study: Just Eat achieves a 3% reduction in bad orders and improved delivery accuracy with Elastic

Just Eat - Customer Case Study

Just Eat, a leading food‑ordering platform with 17.6 million customers and 136.4 million orders in 2016, faced high "bad order" rates caused in part by coarse postcode‑district based delivery rules that allowed orders outside a restaurant’s practical delivery area. The challenge was to make search and delivery radius accuracy more granular and reliable while maintaining fast, stable search performance.

The team implemented Elasticsearch with geospatial polygons (plus Logstash, Kibana and X‑Pack for monitoring and security), converting districts to polygons and using clipped circular delivery zones (via the Clipper library) and KMLs for complex rules; delivery areas now update in near real‑time. The change improved restaurant and customer satisfaction, enabled rapid root‑cause analysis, and reduced the UK bad order rate by 3% overall (with up to 20% reductions at some restaurants).

Just Eat

Matthew Jones

Principal Engineer