Understanding household and person relationships using entity resolution on Snowflake

The Orthodox Union (OU) is one of the largest Orthodox Jewish organizations in the US, founded in 1898. They run 40+ websites, 5 mobile apps, and multiple custom CRMs across departments, each generating separate donor and member records with no shared identifier across systems.

Understanding donors and household relationships across 40+ custom CRMs was impossible without unified data. Person-level and household-level matching both needed to work across decoupled systems without disrupting existing architectures. Initial runtime for the full data load was 12 hours, far too slow for the planned multi-batch-per-day cadence.

The Orthodox Union put Zingg Enterprise at the center of their data architecture on Snowflake, with Airbyte bringing data in from Postgres and MongoDB, and dbt handling cleansing. They built both person-level and household-level matching models, and worked with the Zingg team to develop the Incremental Flow feature that made multi-batch-per-day processing possible.

• Both person-level and household-level matching models built and running in production
• Runtime reduced from 12 hours to 3 hours for initial load, now running incrementally 4–8 times per day
• Family relationship trees constructed automatically from matched household data