CMDB rot is not a one-time cleanup. It is a debt that compounds weekly. The teams I see staying above 90% CSDM compliance run a fixed quarterly cadence, not a yearly heroic sweep.
Week 1: Stale and orphan sweep
Pull every CI with last_discovered older than 45 days that is still operational_status=1. These are either decommissioned-but-not-retired or off the discovery path.
var gr = new GlideRecord('cmdb_ci');
gr.addQuery('operational_status', 1);
gr.addQuery('sys_class_name', '!=', 'cmdb_ci_appl');
gr.addEncodedQuery('last_discoveredRELATIVELT@dayofweek@ago@45');
gr.query();Bucket the output into three lists: confirmed-retire, missing-MID-coverage, manually-maintained. Send each to its owning group with a 10-business-day SLA.
Week 2: Relationship sanity
The relationship table grows faster than CIs. Run the Relationship Audit job and look at the top 20 most-related sys_ids. Any CI with more than 200 inbound relationships is almost always a bad reconciliation rule, not a legitimate hub.
Week 3: Identifier and reconciliation rules
Open the CI Class Manager. For every class above 10K rows, confirm the identifier rules use at least two attributes and that the reconciliation rule list does not include Discovery for fields manually owned by an app team. This is where duplicate CIs are born.
Week 4: CSDM mapping check
Run the CSDM Data Foundations dashboard. For every Business Application missing a Service Offering, send a one-click form to the application owner. Track completion as a single KPI on your steering deck.
The metric that matters
Stop reporting “CI count.” Report % of Business Critical CIs with verified owner, environment, and lifecycle stage in the last 90 days. Executives understand it and discovery teams can move it.
What to do this week
Schedule the four jobs above as recurring sysauto entries with calendar invites for the owning teams. The hygiene problem disappears when the cadence is not a heroic act.
