An AI pilot can save time and still leave the buyer unsure whether it was worth the money. Usually, the missing step is explaining what happens to the time once it is available.
A seller who spends less time preparing can cover more useful conversations. An operations team can absorb more work before it needs another hire. A fixed-fee service can take fewer hours to deliver. Those are different economic paths, and each needs its own evidence.
I start with the task. Time it as people actually perform it, including checking, correcting and putting the result where someone can use it. The model's response time is only one part of that work.
A worked example
Consider a team preparing 30 sales follow-ups per week. The following numbers are illustrative, not a client result. Replace them with timings from your own workflow.
| Measure | Current process | Pilot |
|---|---|---|
| Comparable completed follow-ups per week | 30 | 30 |
| Minutes per task, including review | 24 | 10 |
| Total task minutes | 720 | 300 |
| Additional weekly maintenance | 0 | 90 |
| Total time | 720 minutes | 390 minutes |
The pilot returns 330 minutes, or 5.5 hours each week. The 58% reduction in task time becomes a 46% reduction after maintenance is included. That is why the operating work belongs in the calculation. If you want to value the time, use your own hourly figure for the people doing this work.
If rejected outputs were redone manually, add that time. If the pilot used easier cases, fix the comparison. Record abandoned tasks so that the fastest successes do not become the whole story.
Decide what those hours will do
Suppose the team uses the time for additional qualified customer conversations. Track whether those conversations happen and what comes next. If management intends to defer a hire, record the workload that would otherwise have required it. If the work is billed by the hour, examine write-offs, service quality and additional demand before assuming faster delivery increases revenue.
The same workflow can be valuable for different reasons. A CRO might care most about response speed. A COO might need the same output to require fewer handoffs. Finance will want to understand implementation cost, ongoing subscriptions, maintenance and the route from capacity to an actual financial effect.
Keep four lines in the value record
Operational change: net hours returned on a defined task and cohort.
Use of capacity: the additional work the team actually completed.
Financial effect: an expense avoided or reduced, if one occurred.
Commercial effect: resulting opportunities or revenue, with record IDs.Do not add these lines together as if they were independent benefits. If the same hours support an avoided hire, counting their capacity value again as cash saving would double-count the result. Similarly, pipeline and closed revenue sit at different stages of the same commercial path.
Agree the primary measure before the pilot begins. Keep the quality standard alongside it, because lower effort is useful only if the output remains fit for the work. Review the economics again after normal use, when corrections and maintenance are visible.
The first investment decision becomes easier when the buyer can inspect this calculation and name what changes in the team's week. Bring the recurring task, its frequency and a few timed examples to the first conversation. We can work from there.