ANA-004 recommended Data Pipeline

BI reporting layer

BI tool connected to data warehouse. Non-engineers can view dashboards and ideally create their own reports (self-serve).

Question to ask

"Can your non-engineers answer their own data questions?"

What to check

  • Check for Metabase config or Docker service
  • Look for alternative BI tools (Looker, Superset, Redash, Tableau)
  • Verify non-engineers have access
  • Check for embedded analytics if customer-facing

Verification guide

Severity: Recommended

Data in a warehouse is only valuable if it's accessible. A BI layer lets non-engineers build reports and dashboards without writing SQL.

Check automatically:

  1. Check for Metabase:
# Metabase references
grep -riE "metabase|METABASE" .env.example .env* docker-compose*.yml 2>/dev/null

# Metabase in Docker
grep -iE "metabase" docker-compose*.yml 2>/dev/null
  1. Check for alternative BI tools:
# Looker, Superset, Redash, Tableau, Mode
grep -riE "looker|superset|redash|tableau|mode.*analytics|preset\.io" . 2>/dev/null | grep -v node_modules | head -10

# Looker specifically (often in config or .lookml files)
find . -name "*.lookml" -o -name "*.lkml" 2>/dev/null | head -5
ls -la looker/ lookml/ 2>/dev/null
  1. Check for embedded analytics:
# Embedded dashboard configs
grep -riE "embed.*dashboard|dashboard.*embed|iframe.*metabase|iframe.*looker" --include="*.ts" --include="*.js" --include="*.tsx" . 2>/dev/null | grep -v node_modules | head -5

If not found in code, ask user:

  • "What tool do you use for analytics dashboards and reports?"
  • "Can non-engineers (product, marketing) build their own reports?"
  • "Is there a URL where I can see your analytics dashboards?"

Cross-reference with:

  • ANA-002 (BI connects to the warehouse)

Pass criteria:

  • BI tool connected to data warehouse
  • Non-engineers can view dashboards
  • Ability to create new reports without engineering help (self-serve)

Fail criteria:

  • No BI layer - only raw SQL access to warehouse
  • Reports are one-off scripts engineers run manually
  • No dashboards exist

Partial (acceptable):

  • Dashboards exist but only engineers can create/modify them
  • Read-only access for non-engineers (can view, not build)

Recommendations:

  • Metabase (recommended) - open source, easy to deploy, good SQL mode
  • Looker - enterprise, strong semantic layer
  • Superset - open source alternative
  • Redash - lightweight, query-focused

Evidence to capture:

  • BI tool name
  • Dashboard URL (if shareable)
  • Who has access (engineers only, or self-serve for all)
  • Number of active dashboards/reports

Section

18. Analytics

Performance & Analytics