The task
Scenario. Metadata drifts. In the analytics schema, certifications from last year are still marked current, a few owners have since left, and some descriptions describe a table that has since been reshaped. You want to surface what needs attention before the drift confuses someone downstream.
The ask
- Tables in analytics certified more than 6 months ago
- A review of which descriptions and owners look out of date
- A warning announcement on the ones that need recertification
Hero prompt
Find all tables in the analytics schema certified more than 6 months ago. Review their current descriptions and owners, and flag any that need recertification with a warning announcement.
Step by step
-
Search for assets with stale or incomplete metadata.
Find tables in the <schema> schema certified more than <6 months> ago.
-
Compare what's changed against what's documented.
Show which descriptions or owners look out of date.
-
Update the fields, and announce it if the asset is in active use.
Flag the ones needing recertification with a warning announcement.
What you'll give it
Required context
- The scope to check (a schema, a domain, or the whole estate)
- Your staleness threshold (certified more than 6 months ago)
- Permission to post announcements and update metadata
Optional context
- A definition of "in active use" (queried in the last 30 days, say), so announcements go only where they matter
- The recertification owner to route flags to
What Atlan creates
Outputs
- A list of analytics tables certified more than 6 months ago
- A read on which descriptions and owners look out of date
- A warning announcement on the tables that need recertification
Summary. Drift becomes a short, ranked worklist instead of a surprise. Anyone who opens a flagged table sees the recertification notice in context.
Changed: a warning announcement on the flagged tables (and any field you approved). Left untouched: certifications themselves, which stay as they are until a human recertifies.
Follow-up prompts
Keep going once the first answer lands.
Narrow to what matters
Of the flagged tables, show only the ones queried in the last 30 days (where usage data exists); ignore the dormant ones.
Draft the recert checklist
For each flagged table, list what a reviewer should confirm before recertifying it.
Route it
Group the flags by owning team so I can send each team its own list.
Tips & troubleshooting
Schedule this monthly. Drift is easier to fix a little at a time than all at once.
- Match the tool to the work. This posts announcements; run it where you can see the flagged list before anything is posted, and confirm the scope.
- Automate or chain it. This is the recipe most worth putting on a schedule: a monthly sweep keeps the estate honest without anyone remembering to run it.
- Ask what it changed. Add "list every asset you posted an announcement on." That is your monthly drift log.