Data Quality Monitoring

Critical tables tiered and owned, tests for freshness, volume, duplicates and validity, alerts that reach the right owner, and a data incident process that ends in a fixed root cause.

Free · Mac (Apple silicon) and Windows · v1.0.5 · Other platforms

Then open the Marketplace in the app, search for Data Quality Monitoring, and click Install. It goes into your own files, and works offline from then on.

What it installs.

1

Pillar

1

Area

50

Projects

197

Starter tasks

📊 Data & Analytics

Data Quality Monitoring

Testing data for freshness, completeness, duplicates and anomalies, alerting owners when it breaks and tracking incidents to root cause.

  1. Critical dataset inventory for quality monitoring
  2. Impact tiers for every monitored dataset
  3. Named owner for every tier one and tier two table
  4. Quality expectations gathered from data consumers
  5. Freshness and completeness targets per tier
  6. Baseline quality audit of the ten most used tables
  7. Choosing a data quality testing tool
  8. Failed check alerts routed to the owning team
  9. Severity scale for data quality failures
  10. Data incident log with root cause fields
  11. Morning triage of failed quality checks
  12. Weekly alert noise review
  13. Monthly data incident root cause review
  14. Monthly data quality scorecard
  15. Quarterly test coverage review of tier one tables
  16. Data quality on-call rota and handover
  17. Data status notices for downstream consumers
  18. Quarterly threshold recalibration
  19. Quarantined record review with source teams
  20. Data entry quality feedback to operational teams
  21. Freshness checks that respect load schedules
  22. Volume checks with weekday and seasonal baselines
  23. Duplicate detection beyond the primary key
  24. Column completeness and null rate thresholds
  25. Validity rules for formats, ranges and accepted values
  26. Automated source-to-warehouse reconciliation tests
  27. Blameless root cause analysis for data incidents
  28. Working out the blast radius of a broken table
  29. Write, audit, publish for tier one tables
  30. Circuit breakers that stop bad loads reaching dashboards
  31. Quarantine tables for rejected records
  32. Halving the volume of noisy alerts
  33. Rule-based checks or automated profiling monitors
  34. Data diffs on pull requests before merge
  35. Check results history table in the warehouse
  36. Eliminating the most frequent incident cause
  37. Go-live of the first tier one quality targets
  38. Quality watch through a source system migration
  39. Quarter-end close data quality watch
  40. Data incident fire drill
  41. Quality gate before a new data product launches
  42. Annual data quality review with leadership
  43. Data quality for a data team of one or two
  44. Inheriting a warehouse with no tests
  45. Suspected data issue guide for analysts
  46. Onboarding an engineer onto data quality on-call
  47. Quality controls for regulated or audited reporting
  48. Data quality SLOs with error budgets
  49. Sampled accuracy audits against source records
  50. Quality checks on product event tracking data

Where it starts.

Each project arrives with its purpose, milestones and first tasks written out. Here is how the first three read.

Critical dataset inventory for quality monitoring

This is a team plan, best installed into a shared workspace so everyone with a role in it can see and own their tasks. You cannot monitor everything, and teams that try end up with thousands of checks and no idea which failures matter. Listing the tables and files that feed revenue reporting, customer-facing features, regulatory returns and the most used dashboards gives a short, defensible list to protect first.

First tasks

  • Export ninety days of query history from the warehouse usage views
  • List the tables behind the twenty most viewed dashboards
  • Add tables that feed product features, exports and reverse syncs
  • Ask each consumer team lead to confirm or challenge the top list

Impact tiers for every monitored dataset

Tiering decides how fast a failure gets attention, how strict the checks are and who gets woken up, so it should be a written decision rather than a feeling during an incident. Three tiers work for most teams: tier one breaks money, customers or compliance, tier two breaks a team's daily work, and tier three is exploratory.

First tasks

  • Draft one-paragraph criteria for each of the three tiers
  • Assign a tier and a reason to every dataset on the inventory
  • Tag each table with its tier in metadata or the model config
  • Re-tier datasets whose consumers or uses have changed @recurring(yearly)

Named owner for every tier one and tier two table

An alert sent to a shared channel with no named owner is read by everyone and acted on by no one. Each important table needs a technical owner who fixes it and a business contact who can say whether a strange value is actually wrong, written down before the first incident rather than negotiated during it.

First tasks

  • List the tier one and tier two tables with no clear owner today
  • Propose an owner for each and ask their manager to confirm
  • Record owner, backup and business contact in table metadata
  • Check owners and backups are still in post after team changes @recurring(quarterly)

Make it yours.

Install Data Quality Monitoring, then tell Aurum the date you're aiming for. It checks whether the plan fits the weeks you actually have.

Free · Mac (Apple silicon) and Windows · v1.0.5 · Other platforms

Blueprints are general information to help you organise your own work. They are not medical, financial, legal or other professional advice. See the Terms.