An inspection program is judged by one conversion rate: how reliably an observed anomaly becomes either a closed evaluation or a completed repair. Everything else, coverage percentages, reports filed, photos taken, is input. The conversion is the output, and at most sites it leaks.
The leak is rarely dramatic. A finding gets written up, discussed in a morning meeting, and everyone assumes someone raised the notification. Fourteen months later the same inspector finds the same defect, slightly worse, with no record that anyone decided anything.
The pipeline, named
Triage works when every finding moves through explicit states, each with an owner. The states can be simple; they must be real.
- Captured. The anomaly exists as a record against the equipment: photos, dimensions, location, severity estimate from the finder. Capture happens in the field, the same shift, or details start decaying immediately.
- Screened. A qualified reviewer confirms severity and classification. This is where the trivial gets closed quickly, and quick closure of trivia is what keeps the queue credible. A queue full of six-month-old cosmetic findings trains everyone to ignore it.
- Evaluated. The finding that survives screening gets an engineering disposition: fit for continued service, monitor at a defined interval, repair by a defined date, or restrict operation. The disposition is recorded with its basis, not just its conclusion.
- Actioned. Monitor becomes a scheduled inspection task. Repair becomes a work order in the maintenance system, carrying the finding's evidence with it. The link between finding and work order is a hard reference in both directions, not a matching description.
- Verified and closed. The repair is confirmed, often with follow-up NDE, and the closure lands back on the finding. Closed means verified, not "work order status says complete."
Where pipelines actually break
- The screening bottleneck. One senior inspector is the gate, and when they are on a turnaround the queue silently grows. Fix: a defined backup and a visible queue age metric.
- The handoff to the CMMS. The finding lives in the inspection system, the work order in the maintenance system, and the connection is a copy-pasted description. When either side changes, the link dissolves. Fix: integration that carries the finding reference into the work order and the completion status back.
- Severity inflation and deflation. With no shared severity definitions, one inspector's critical is another's monitor. Calibrate with written criteria and periodic joint reviews of a sample.
- The disposition that never lands. Engineering says fit for service in a meeting, and the finding stays open forever because nobody wrote it down. Every disposition needs a record and a signature, even the benign ones. Especially the benign ones; they are the ones an incident review will ask about.
The metric that keeps it honest
Track open findings by age and severity, reviewed somewhere it cannot be ignored. Not as a stick, but because an aging critical finding is either a resourcing problem or a decision nobody wants to make, and both deserve daylight. The companion metric is loop closure: what fraction of repair dispositions have a verified completed work order. That number is your real conversion rate.
A finding you cannot trace to a disposition is a liability wearing the costume of diligence.
In intelliSPEC, findings captured in intelliINSPECT move through screening and disposition with the evidence attached, and the CMMS integration carries the work order link both ways, so the record of anomaly to closure assembles as the work happens. Trace one of your own stuck findings at a demo.
