Count the Return Trip, Not Just the Finished Item

Thirty completed items may represent thirty straightforward journeys or thirty journeys plus a pile of corrections. A completion total alone cannot reveal how much work had to be done again. To understand rework, follow the return trip as well as the final destination.

Be careful with the word. A required review is not rework merely because it examines something already created. Nor is an intentional revision after a new request necessarily a defect. The reason work repeats determines what the repeat can tell you.

One fictional print exercise

Fictional example: An office prepares thirty generic dividers for an imaginary reference binder. Each divider must match the size and orientation specified on its invented request. Seven dividers need correction before completion. Five need one correction; two need two corrections.

That gives thirty completed dividers, seven affected dividers, and nine correction episodes. The affected-item proportion is seven out of thirty, about 23 percent. There are nine correction episodes per thirty dividers, or 0.3 per divider. Those are different descriptions, not competing calculations.

If someone reports “nine failures,” a reader cannot tell whether nine separate dividers were affected or some dividers returned more than once. A useful count names its unit: items, episodes, pages reprinted, or minutes of repeated work. Changing the unit changes the question.

Separate four reasons for repeating a step

1. Correction: The prepared item did not meet the accepted requirement
2. Changed request: The requester changed the desired output after the requirement was accepted
3. Planned iteration: Several drafts were always part of the agreed work
4. Unclear requirement: The people involved interpreted the input differently

These categories are editorial choices for this exercise, not a universal coding system. Their value is that they stop all repeated activity from being treated as one problem. An authorized local project would need definitions that fit its own purpose, with uncertainty preserved when the reason is not known.

Frequency is only one view

Suppose incorrect orientation causes six correction episodes, a size mismatch causes two, and an unreadable print causes one. Orientation is the most frequent category. But if each size mismatch takes much longer to correct, the most common category may not account for the most effort.

IHI’s Pareto Chart page describes a display that helps identify the categories contributing most to a total. Decide what that total means before ranking: counts, cost, or time. Our fictional exercise does not establish a universal “80 percent” rule, and a low-frequency issue is not automatically unimportant.

For a real concern involving safety, privacy, or a required control, use the appropriate reporting route regardless of where it sits in a frequency ranking. A process-improvement exercise should never delay an urgent report.

A plausible cause is still a hypothesis

It would be easy to blame the fictional preparer for the orientation errors. Another possibility is that the request shows orientation ambiguously. A third is that the default print setting conflicts with the template. The count alone cannot select among those explanations.

IHI’s Cause and Effect Diagram resource presents a way to organize possible causes. “Possible” matters. A neatly organized diagram is not evidence that any particular explanation is true. In this exercise, write the competing explanations and what a permitted, fictional test would need to distinguish them.

Look for repeat demand that escapes the main count

A corrected divider may also trigger a new check, another transfer, and a search for the earlier request. Those activities can be invisible if the only measure is reprinting. Draw the full loop on a process map before claiming that one minute of correction represents the entire burden.

Stay with invented data for practice. Do not save real rejected forms, employee performance details, or patient information to build a personal rework log. Any operational review needs approved definitions, access, and ownership. The goal of this article is to improve the question being asked: how often does work return, why does it return, and what other work follows it?

Evidence boundary: The two IHI public resource pages were checked October 8, 2026. All divider counts, classifications, and interpretations are original teaching examples, not results from LIJ, Northwell, or an IHI case study.