Data Warehouse

A data warehouse is a centralized system designed to store, organize, and query large volumes of structured data from multiple sources. Unlike a transactional database (which handles real-time reads and writes for your application), a data warehouse is optimized for analytical queries: aggregations, joins across datasets, historical trend analysis, and reporting.

Why it matters: as a company grows, data fragments across dozens of tools. Your product usage data lives in your application database. Marketing data sits in Google Analytics, HubSpot, and ad platforms. Revenue data is in Stripe and your CRM. Support data is in Zendesk. Without a warehouse, answering questions like "what is the LTV of customers acquired through organic search who use feature X within their first week?" requires manually pulling and joining data from four different systems. A warehouse centralizes everything into one queryable location.

The major platforms: Snowflake, Google BigQuery, Amazon Redshift, and Databricks are the leading cloud data warehouses. BigQuery is popular for its serverless model (no infrastructure management) and tight integration with Google's ecosystem. Snowflake dominates in the enterprise space with its separation of compute and storage. For startups, BigQuery or even a managed Postgres instance can be sufficient.

How data gets there: ETL (Extract, Transform, Load) or ELT (Extract, Load, Transform) pipelines move data from source systems into the warehouse. Tools like Fivetran, Airbyte, and Stitch handle the extraction and loading. dbt (data build tool) has become the standard for transforming data once it lands in the warehouse, letting analysts write SQL models that clean, join, and structure raw data into analytics-ready tables.

Common mistakes: treating the warehouse as a dumping ground without governance. If you load data without a clear schema, naming conventions, or documentation, you end up with a data swamp. Also, underestimating the cost of compute-heavy queries on platforms like BigQuery or Snowflake, where poorly written queries can rack up significant bills.

Practical example: a SaaS company connects Stripe billing data, HubSpot CRM data, and Mixpanel product usage data in BigQuery using Fivetran. With dbt, they build a unified customer table that joins subscription revenue with feature usage and acquisition source. Now their growth team can run queries like "show me churn rate by acquisition channel and onboarding completion status" in seconds.