In this article
A process can improve on the number being watched while becoming harder to live with. A faster first step may flood the second step. A shorter form may create more follow-up work. A quicker completion label may leave essential finishing work outside the clock.
IHI’s Establishing Measures page distinguishes outcome, process, and balancing measures. The balancing perspective asks whether a change creates problems elsewhere in the system. The examples below apply that idea to invented office work; they are not recommended Northwell metrics or targets.
Start with how the harm could happen
Fictional example: An office prepares generic packets for an imaginary lecture. A proposed arrangement sends each packet to checking immediately instead of waiting for a group. The intended benefit is earlier readiness. A plausible downside is that frequent transfers interrupt the checker’s other scheduled desk exercise.
“Could anything go wrong?” is too broad to produce a useful measurement plan. A more concrete question is: “Does the transfer change increase the number of times the checking task is interrupted or restarted?” The mechanism links the proposed change to the concern.
Another possible downside is extra handling. If the exercise assumes transfers are instantaneous and effortless, it cannot answer that question. Recognizing the missing mechanism is an important result. It tells you where a model ends and an authorized operational investigation would have to begin.
Keep the three views distinct
- Outcome view: When are all required packets checked and ready?
- Process view: Are packets transferred according to the agreed mock-test rule?
- Balancing view: Does the new transfer arrangement create additional handling or disruption?
The labels depend on the project’s question. A measure that serves as a balance in one project may be the primary outcome in another. Do not spend more effort debating labels than explaining what each measure protects against.
Pair a gain with its nearest credible downside
Consider a second invented proposal: remove an optional formatting step from a generic notice. The intended gain is less preparation time. If the removed formatting helped readers distinguish sections, the credible downside might be reduced usability. Counting only preparation minutes would miss that effect.
That does not mean every optional feature is valuable. It means a claim of improvement should preserve the purpose of the output. Before any real change, the appropriate owner would need to determine whether the formatting was actually optional and whether accessibility or other requirements apply. A fictional exercise cannot answer those questions.
Agree on the response before seeing the result
Suppose the packet exercise becomes faster but handling effort rises. Is that acceptable? The arithmetic cannot make the decision. The responsible people need to decide what matters, what tradeoffs are permissible, and what result would stop or revise a test.
Do not borrow an arbitrary threshold from an article. A five-minute burden may be trivial in one invented model and consequential in another real setting. Clinical, safety, privacy, accessibility, labor, and service requirements are not optional tradeoffs to be optimized by a personal experiment.
An authorized plan should also say who reviews an unexpected effect and how the previous arrangement can be restored if appropriate. “We will watch it” is not a complete response plan if nobody has responsibility for interpreting the result.
A balance measure should not become another burden
A long list of measures can consume time and produce little useful evidence. For the fictional packet exercise, one or two carefully chosen counterchecks may teach more than a sprawling dashboard. For real work, the choice of measures and any data collection must be approved, proportionate, and supported by available evidence.
Do not monitor colleagues, record personal work patterns, or copy patient or workforce data to create a balancing measure yourself. The constructive next step is to describe the plausible downstream effect to the process owner and ask how an authorized project would assess it.
The sentence to complete
“We expect this change to improve this outcome, but it might make this other part harder because of this mechanism.” If you cannot complete the last clause, explore the process before choosing a convenient metric. If you can, you have a sharper question for a permission-bounded small test.
Source scope: IHI’s public measurement guidance was checked October 8, 2026. Its broad measurement categories support this discussion; the packet examples, tradeoff questions, and permission guidance are original editorial applications, not employer policy.