Measure the whole job

The method

1. Choose comparable cases

   Measure the same task type, similar difficulty and the same completion point. Include incomplete inputs and a typical exception in the trial.

2. Add up all the effort

   Preparation + processing + review + correction. Record waiting separately. State whether a figure is measured or estimated and how many cases it represents.

3. Put quality beside speed

   How many results passed on the first attempt? What was the consequence of each error? Faster processing is useful only if the agreed quality conditions are met.

The rework behind a fast answer

Illustrative example using fictional trial data. The old method takes 10 minutes of preparation, 12 of processing, 4 of review and 3 of correction per case. The new figures are 5, 4, 8 and 6 minutes.

Total effort: 29 → 23 minutes, a difference of 6 minutes per case. Processing is 8 minutes shorter, but review and correction together rise from 7 to 14 minutes. First-pass acceptance across ten cases: old method 9, new method 8. Decision: fix the quality issue and repeat the trial; the time result alone does not justify expansion.

Your worksheet

Trial boundary

Which tasks, sample size and common completion point will you compare?

Total effort

Preparation, processing, review and correction: before and after. Record waiting time separately.

Quality and exceptions

What counts as an acceptable output? How many cases succeeded, failed or needed correction?

Decision after the trial

Which conditions trigger expansion, revision or rollback? Who decides, and when will you measure again?

Review checks

• Both methods were measured through the same complete task.
• The time result includes its sample size and quality figures.
• The causes of failed cases were examined before deciding what happens next.

Make correction time visible

For the next five eligible cases, time review and correction separately. Without these, the apparent improvement can be incomplete.