Insights

What to measure after a workflow goes live

2026-11-03 · 5 minute read

Count cycle time, exceptions, rework, and the share of volume on the new path. Do not stop at a task counter.

Automation projects collect the wrong scoreboard. "Tasks automated" goes up while coordinators quietly re-do the work, and the leadership meeting still hears that intake is behind. The measure has to describe the operation, not the tool.

A small set that holds up

Cycle time from a defined start to a defined finish. For referrals, that might be packet arrival to a scheduled start of care. For scheduling, patient ready to appointment confirmed. Write the definition down before the build, and use the same definition after.

The share of volume that actually travels the new path. A workflow that handles the easy third and leaves the rest in the old inbox has not been implemented.

Exception rate and how long exceptions sit. Rework after an automated step. Hours returned only where you can point to tasks people no longer perform. If the organization cares about administrative leakage, track the specific failure: delayed admission, missing authorization handoff, documentation that has to be chased. Do not invent a dollar figure the finance lead cannot trace.

Baseline, then resist the anecdote

Without a baseline, every later number is a story. Pull the baseline from the current process, even if it is ugly. After go-live, review the same numbers with the operator who owns the queue, not only with the project sponsor.

Published results from someone's previous role are context for how they think about the work. They are not a forecast for yours. Umble treats founder outcomes that way on purpose, and an engagement should earn its own measures.