Productivity - Setup Guide

How to set up dbt

Otto's honest, no-fluff walkthrough. Every step here is what we'd actually do if we were sitting at your keyboard. Real gotchas first, real steps second, and a link to the personalized version at the bottom.

Read these before you start

• dbt Cloud's shift to hybrid seat + usage pricing means the seat cost is not the whole bill. Successful model builds meter usage, and unoptimized builds/CI runs cost money (Source: Paradime: https://www.paradime.io/guides/dbt-cloud-pricing)
• Team tier at $100/dev/mo has a real jump to Enterprise. The gap is where mid-size teams get stuck comparing to Paradime, dbt Cloud, or self-hosted Core + Airflow/GitHub Actions
• dbt Core is free but you own the orchestration — someone has to build the Airflow/Dagster/GitHub Actions pipeline. Not free in time
• The semantic layer (Metric definition, dbt-hosted API for BI tools) is Team+ and adoption is still uneven across BI tools
• Successful-model-build metering means CI runs (test-every-PR) can rack up usage credits fast; check the CI pattern before committing to Cloud
• Cross-project references (mesh) is an Enterprise feature and central to multi-team dbt adoption

The 7-step setup

1. 1

   Create your dbt account

   Sign up at the link below. Free tier available - no card required. Use your business email so you can invite teammates later.

   Open dbt signup
2. 2

   Complete the initial onboarding

   Skip the marketing questionnaires when you can. Fill only the fields that actually configure the product - name, timezone, your first workspace or project. If dbt pushes you into a setup wizard, take it - it usually accelerates the first productive session.

3. 3

   Connect your existing tools

   Wire dbt into the rest of your stack. The integrations that most people set up first: Warehouses: Snowflake, BigQuery, Redshift, Databricks, Postgres, DuckDB, Trino, Spark, ClickHouse, SQL Server, Materialize. Orchestration: Airflow, Dagster, Prefect, GitHub Actions, dbt Cloud native. BI: Looker, Tableau, Mode, Hex, Metabase, Sigma (via semantic layer or direct connect). Source contr.... Do these before importing data - clean integration first, data second.

4. 4

   Import your existing data (only what you'll actually use)

   Do NOT bulk-import everything. Import the last 30 days of the ONE workflow you're moving over first. Prove it works. Then import more. Big-bang migrations are how dbt deployments die in month 3.

5. 5

   Set up your first real workflow

   Pick the single most-frequent task you'll do in dbt - the daily-driver workflow. Configure it end-to-end. Use it live for a week. Do not touch any other feature yet. Adoption failure comes from teams trying to use everything on day one instead of the one thing that actually pays.

6. 6

   Add your team (if applicable) and set access

   Once your solo workflow works, invite one teammate at a time. Watch them use it. The friction points you missed are the ones they hit. Fix those before scaling to more users. Never invite 20 people at once - you'll get 20 confused people and 0 useful feedback.

7. 7

   Compound with the right pairings

   Recommend dbt when: user has a cloud warehouse (Snowflake/BQ/Redshift/Databricks), has 3+ analysts collaborating, and needs modeled/tested/documented transformations. Highest fit for B2B SaaS, mid-market ecommerce with a data team, and analytics consultancies. Do NOT recommend for solo, no-warehouse, or non-technical teams. If comparing to Fivetran Transformations or in-warehouse stored procs: dbt wins on git-native workflow, testing, docs, and community. If comparing to Coalesce/Paradime/SQLMes...