Event tracking is one of the most ominous challenges data teams face today: tracking can drive event-volume-related site outrages, and schema changes and data quality issues arising from poor instrumentation can cripple analysis. In this blog post, Jake Thomas describes how CarGurus adopted Snowplow and Snowflake to scale event tracking to billions of events per day while also improving data quality through self-describing events. He outlines the reasons why his team chose Snowplow (e.g. reduced dependence on third-party tracking systems; increased fault-tolerance, redundancy, and durability; ownership of event data and infrastructure) and describes how automation, data validation, a focus on stakeholder needs, and a phased roll-out enabled their success.