Case Study: Marriott achieves dramatic pricing performance — from 4+ minutes to ~13 seconds — with Neo4j

Marriott - Customer Case Study

Marriott International faced slow, resource-intensive pricing updates for searches across its ~4,500 properties: using a relational database to compute seasonal rules, rate modifiers and “touch one, modify many” pricing meant some publishing jobs took minutes (and sometimes hours), generating as many as 30,000 SQL calls and causing backlogs that hurt the business.

Marriott implemented Neo4j as a cache layer (keeping Oracle as the system of record), modeled relationships as first‑class graph entities, and added custom sharding across clusters so each property’s graph is processed in memory. The change cut average processing from over four minutes to about 13 seconds, drove a 300% increase in price-change volume, reduced CPU needs by an order of magnitude, and cut infrastructure costs roughly 50%, while improving adoption and responsiveness.

Marriott

Scott Grimes

Senior Director