The temptation is to build a data warehouse first
Series A companies often hire their first data person and immediately start evaluating Snowflake, dbt, Fivetran, and Looker. This is almost always wrong.
Before you invest in infrastructure, you need to know what decisions you're trying to make. Infrastructure without questions is just expensive storage.
What you need in the first 90 days
The minimal viable data stack for a B2B SaaS company looks like this:
- A single source of truth for revenue: your CRM and billing system are reconciled and trusted. Nothing else matters until this works.
- One dashboard that answers three questions: What is ARR today? Where is churn coming from? What does the pipeline look like?
- A data dictionary (even a Google Doc) that defines your key metrics in writing.
That's it. You don't need a warehouse yet.
When to add infrastructure
Add a data warehouse when you hit one of these conditions:
- You're making decisions from more than 3 different data sources
- Your BI tool is timing out on queries
- You have more than one person who needs data access
Until then, stay in the tools you already pay for.
The real cost of moving too fast
The companies that over-build early spend 6 months integrating tools instead of building understanding. By the time the warehouse is running, the business has changed and the initial data model is already wrong.
Start with questions. Build infrastructure to answer them. That order matters.