Quickstart

This guide takes you from zero to a validated data asset: connect a source, profile an asset, let AI draft rules, group them, and run.

Time to complete: ~5 minutes You'll need: a Measured account (or a self-hosted instance) and a warehouse/database to point at.

1. Get Measured

  • Cloud (fastest): sign up at app.measured.cloud and create your organization.
  • Self-host (Enterprise): run the container on your own infrastructure.
docker run -d --name measured -p 8080:8080 \
  -e DATABASE_URL="postgres://user:pass@your-db-host:5432/measured" \
  -e SECRET_KEY="change-me-in-production" \
  registry.measured.cloud/measured:latest

Open the app and complete the setup wizard to create your admin account.

2. Connect a data source

Data Sources → Add data source. Pick a type (Snowflake, BigQuery, Postgres, Databricks, Oracle, and more), paste a connection string, and Test it. On Enterprise, route the connection through a Privacy Agent so credentials never leave your network.

3. Discover & profile an asset

Open the data source and Discover its tables. Promote the ones you care about to data assets, then Profile an asset. Profiling captures column stats, completeness, and structure, the signal Measured's AI uses to understand your data and the basis for the asset's health score.

4. Add rules, let AI draft them

Open the data asset. Rules live on the asset. Click Suggest rules and Measured drafts expectations from the profile, completeness, ranges, allowed values, uniqueness, and more. Review, edit, and add your own:

  • Column must not be null, e.g. security_id
  • Column value between, e.g. market_value_usd ≥ 0
  • Column in set, e.g. asset_class in {Equity, Fixed Income, Cash}

Save the rules on the asset.

5. Build a rule group

A rule group is the runnable, health-scored unit. Rule Groups → New rule group, then add members, rules from this asset (and others, if you want one group to span multiple assets).

6. Run it

Run now on the rule group (or trigger it from the API, a schedule, or an agent over MCP). Each run produces a result with pass/fail per rule, failure counts, and an AI root-cause summary, and updates the group's health score.

7. Schedule & act

Attach a cron schedule to run automatically:

0 */6 * * *   # every 6 hours
0 8 * * 1-5   # weekdays at 8am UTC

Then add remediation, alerts (Slack, PagerDuty, webhook, …) or an auto-created incident, so a failure turns into action, not just a red mark.

Next steps