Is this a 40% productivity gain?
Count accepted work and review time before recommending a rollout.
The question
Cairn Operations runs an AI briefing pilot for a fictional German industrial supplier. Two teams each prepare 20 comparable briefs.
| Manual team | AI-assisted team | |
|---|---|---|
| Briefs submitted | 20 | 20 |
| Briefs accepted without material rework | 18 | 16 |
| Preparation hours | 40 | 24 |
| Review and rework hours | 10 | 18 |
Illustrative matched workload. No claim of statistical significance is made.
Preparation time fell from 40 to 24 hours, a 40% productivity improvement. Roll out to every team next week.
“What does this pilot show? Give the sponsor a recommendation and a better success measure.”
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: Check the boundary
What would you do here? Open the move you’d make.
Use 40%; review belongs to a different department.
Tempting, but it misses something. That shifts cost rather than measuring the complete task.
Include review and rework before stating the saving.
The strong move. This compares the work needed to produce the deliverable.
See how a strong candidate says it
“The 40% reduction applies to preparation hours. Total preparation plus review fell from 50 to 42 hours, a 16% reduction. That is the first correction.”
Why it works: Review effort is part of the workflow even when another team pays it.
Step 2: Bring quality into the measure
“Acceptance without material rework fell from 90% to 80%. Total hours per brief accepted without material rework were about 2.78 manually and 2.63 with AI, about 5.5% lower. We do not yet know the final acceptance rate after rework.”
Why it works: A quality-adjusted output measure changes the denominator. It must not be confused with final acceptance after rework.
Step 3: Locate the failure
“Review rejected briefs by error type. Check whether wrong specifications or missing evidence caused the extra work. Do not ask reviewers simply to approve faster.”
Why it works: A quality problem needs diagnosis, not a target that hides it.
Step 4: Recommend a bounded next trial
“Keep the pilot small. Add source-linked specifications and a pre-review check. On another matched batch, track total hours, acceptance before and after rework, material errors and all-in cost.”
Why it works: A second test should change the process and answer the specific uncertainty.
The answer, in short
- Decision
- Do not claim a 40% end-to-end gain or roll out yet.
- Because
- The broader time reduction is 16%, and fewer briefs passed without material rework.
- What would change it
- Expand if a repeatable matched trial improves accepted output per total hour without worse critical errors.
Common traps
- Adding percentages with different denominators.
- Ignoring review labor or tool cost.
Check your answer
- Did I include all 50 versus 42 hours?
- Did I calculate both acceptance rates?
- Did I use accepted output as a denominator?
- Did I give a specific next test?
Last reviewed Nov 10, 2026