Snowplow was founded in 2012 by Alexander Dean and Yali Sassoon in London. The two co-founders came from the analytics consulting world and saw that businesses needed granular, high-quality behavioral data pipelines they could actually own and control.
The platform operates as a behavioral data pipeline. It collects event data from websites, mobile apps, servers, and IoT devices, validates each event against user-defined schemas, enriches the data with additional context (geolocation, campaign attribution, weather data), and delivers clean, structured events to your data warehouse or data lake.
Schema validation is Snowplow’s defining technical feature. Every event must conform to a JSON schema before it’s accepted into the pipeline. Events that don’t match get routed to a “bad rows” queue for inspection. This approach catches data quality issues at the point of collection rather than letting bad data contaminate your warehouse and then trying to clean it up later.
The open-source version (Snowplow Community Edition) handles the core pipeline — collection, validation, enrichment, and loading. Snowplow BDP (Behavioral Data Platform) is the commercial offering that adds managed infrastructure, a UI for pipeline management, and data modeling features.
Snowplow loads data into Snowflake, BigQuery, Redshift, Databricks, and S3/GCS data lakes. Because you own the infrastructure and the data, there are no vendor-imposed limits on data retention or query access. This matters for companies that need complete event histories for machine learning models or regulatory compliance.
The platform has found its strongest adoption among data-mature organizations that have outgrown packaged analytics tools. Companies like Strava, Auto Trader UK, and The Globe and Mail run Snowplow to power their analytics and data science workflows.
For teams that treat behavioral data as critical infrastructure rather than a side feature of their analytics tool, Snowplow provides a level of control and data quality that SaaS analytics platforms can’t match.
