Most CMDB initiatives fail for a simple reason:
They treat data ingestion as success, and data usage as someone else’s problem.
If no team is accountable for consumption, your CMDB is already dead.
Platform PMs live in the uncomfortable middle:
Too technical for strategy decks.
Too strategic for ticket queues.
If you’re comfortable, you’re probably not close enough to the system.
Most infra roadmaps are feature lists.
Real platform roadmaps answer:
- What becomes simpler?
- What becomes more predictable?
- What stops breaking silently?
That’s harder work. That’s the job.
@mikesaleme Exactly.
Additive work ships fast.
Subtractive work scales quietly.
Most platform failures happen when people start optimizing for delivery velocity, not total system cost.
Internal platforms don’t need more features.
They need fewer exceptions.
Every exception:
Increases cognitive load
Creates tribal knowledge
Reduces trust
Platform PM work is subtractive, not additive.
“Just onboard another discovery source” is rarely neutral.
Every new source:
- Competes for truth
- Introduces drift
- Adds reconciliation debt
Scale isn’t free. It’s deferred complexity.
The fastest way to kill CMDB trust:
Change reconciliation rules without telling consumers.
Second fastest:
Let multiple teams “temporarily” override data.
Both look harmless.
Neither is.
Automation isn’t leverage. Alignment is.
You can automate:
- Syncs
- Jobs
- Pipelines
But you can’t automate:
- Ownership,
- Accountability
- Trade-offs
Most platform failures start there.