Picture a timer stopping when the last row appears. The table is neatly filled. Then someone opens the source files, finds a wrong pack size and spends twenty minutes repairing the comparison. That time belongs to the job too.
We would settle the finish line before trying an assistant on a purchasing task. Otherwise the demonstration can end just before the work you most need to measure.
Take a made-up rehearsal with ten quote rows. Manual preparation, including the checks needed to make every row acceptable, takes 50 minutes. Assisted preparation takes 15 minutes. Reviewing that output takes another 25. Fixing the errors found during review takes 20 more. Those are separate, non-overlapping periods in this fictional example.
The assisted route takes 60 minutes of operator work: 15 + 25 + 20. The manual route takes 50. That is six minutes per accepted row rather than five, provided all ten rows actually meet the same acceptance rule. If some remain unresolved, retain their time and report the accepted count separately.
No software was timed here. The numbers illustrate what belongs in the total. They do not establish that an assistant is slower, or that a particular product is faster. A real rehearsal needs its own observations.
Before starting, write down what an acceptable row contains. For this example, it needs the original quote reference, pack size, quantity, confirmed charges and any missing information. An empty freight amount may be correct when the quote never supplied it. A confident guess should not pass because it makes the total look finished.
Here is a small timing record to copy:
Task and finish rule:
Number attempted:
Preparation time:
Review time:
Correction time:
Number accepted at the finish:
Cases still unresolved:
Keep manual and assisted work comparable. If the manual timing already includes checking, do not add the same review period again. If the assisted version leaves a person doing the last checks, do not stop its clock at draft creation. Record waiting separately from active operator time if both matter to the decision.
The next experiment might use new but similar rows, so remembering the first answers does not become the apparent benefit. Keep the same finish rule. A ten-row rehearsal can uncover a troublesome error; it cannot by itself establish the economics of a whole operation.
Our freight example gives you a useful awkward row to include: the goods price is known and delivery is missing.
Read the freight example
For the next task you time, write the acceptance rule above the stopwatch record. Record preparation, review and correction through the work needed to meet that rule. Keep waiting time separate from active work. Then inspect where the minutes went before deciding what to automate next.
The Rebuild Editorial


