The Number Everyone Trusted Was Wrong by Days

The problem

A team reviewing compliance cases wanted a recurring report on individual reviewer throughput and turnaround time — a basic accountability question that had never been measured systematically.

The question

Before building the report: was the data already being tracked actually measuring what everyone assumed it measured?

The approach

Rather than trusting the obvious field labeled “response time,” the underlying event history was reconstructed for a sample of cases and compared against what the field actually reported. Each case’s true trail of events was pieced together from log tables rather than relying on a single status field, and turnaround and closure metrics were explicitly defined for two different request types — including how to handle a reviewer with zero completed cases in a period, so nobody would silently disappear from the report.

What I found

The field everyone had been using to measure response time was capturing the wrong event. It had been quietly understating real turnaround time by a meaningful margin — in some cases by days — for months, without anyone noticing.

What changed

The corrected metric became the basis for a new monthly reporting pipeline, automated to regenerate itself without manual date entry each cycle. More importantly, it reset the team’s baseline expectation of what “normal” turnaround actually looked like.