What a Data Warehouse Looks Like
A data warehouse is five pieces, and you already have the first one.
It's one place that pulls in a copy of what your systems already know, so every report comes from the same set of numbers.
The systems you already run
Your accounting software, your online store, your ERP, your spreadsheets. None of it gets replaced.
Something that moves the data
It runs on a schedule and copies each system into one place, so nobody exports anything by hand.
The warehouse
One database that holds all of it together. This is the part people mean when they say data warehouse.
A layer that defines your terms
Where margin, active customer, and on time get one definition your whole company uses.
Dashboards
What your team opens in the morning. Charts, numbers, and a way to dig into them.
Same five pieces, different tools.
The first is where I'd start most businesses. The other three are for when something specific about your data, your risk, or your team changes the answer.
Where I'd start you
from about $250/month*The general purpose build, with costs you can predict. Snowflake sleeps when nobody's querying it so you pay for the hours you use, dbt Core is free, and Preset is a flat monthly price.
Most of your data comes from Google
from about $250/month*Google Ads, Google Analytics, Search Console. Those move into BigQuery with the least work, and Looker Studio is free.
Regulated data, or it can't go down
from about $1,000/month*Snowflake Business Critical adds failover to a second region, encryption keys you hold yourself, and HIPAA and PCI coverage. Row and column-level security starts one tier below it, at Enterprise. The cost reflects that tier, plus the support and service levels you'll want from every tool in the stack.
You want to own it, and you have a team
from about $75/month*PostgreSQL, Python doing the ELT, dbt Core, and Superset, all on servers you control, plus a local Superset MCP so you can ask questions of it from an AI assistant. The software is free. What it costs you is someone keeping it running.
Already on AWS, Azure, or Google Cloud? Snowflake runs on all three and can bill through that cloud's marketplace, so being on one cloud doesn't mean using that cloud's own warehouse.
If reporting isn't the main job and you're training and running models against your own numbers, that's a different kind of platform: Databricks, or Snowflake with Cortex, where the models and the data sit in the same place.
Ready to see it on your own data first? Back to Data Warehouse
Want to know which one fits you?
Tell me what you're running today and what you're trying to see. I'll tell you which setup fits and roughly what it costs for your situation.
*Note: These are platform costs only. They don't include my time to build it, or your team's time to keep a self-hosted setup running. Estimates assume daily use at entry-level pricing, and your actual cost depends on how much data you have and how heavily your team uses it.
Platform pricing and features as of September 2026.