Audits that changed decisions
Before building anything new, I checked the data we already had. Each audit was small and answered one question. Together they surfaced risks nobody had seen, and each one ended with a decision rather than a report.
The problem
The team trusted its data because it was in a system. But nobody had checked whether the systems matched reality, and new tools were about to be built on top of them.
Constraints
No time set aside
Audits had to fit around daily work, so each one had to stay small.
Read-only first
I looked without changing anything, then brought what I found to the right person.
Many systems
The answers were spread across several tools that didn't always agree.
Key decisions
Ask one question per audit.
For example, are the records that say "current" actually current? Small questions got answered, where a big clean-up would have stalled.
Show a few real examples.
Every finding came with examples, so the person responsible could see it for themselves.
Take findings to the owner.
I didn't quietly fix records. Each finding went to the person responsible, so the cause got fixed, not just the symptom.
How it works
What changed
- Compliance records that looked current but weren't were caught, and the maintenance team got a dashboard showing what was actually up to date.
- A feedback survey question that had never collected an answer was found and fixed.
- Inconsistent records were flagged for clean-up before any new tool relied on them.
- Every automation was reviewed, and the broken and duplicate ones were flagged.
What I'd do differently
- Schedule audits instead of running them when something feels off. A regular check on the riskiest data would catch problems sooner.