1. Agree the baseline together
Pick a representative week before the pilot. Count the in-scope items and time the work. Use the team’s actual records, not a generic industry average. Note unusual workloads so you do not compare a quiet week with a busy one.
2. Count the review work
Net time returned is the previous handling time minus the time still spent reviewing, correcting and managing the new workflow. If preparation gets faster but checking takes longer, record both. AI activity counts are useful operating data, not proof of savings.
3. Keep quality beside speed
Track correctly handled items, factual edits, missed escalations and any unapproved actions. Define a sample and an acceptance rule before the pilot. A fast draft with the wrong amount or date is not a successful outcome.
4. Translate time carefully
If you choose to value time in euros, use an agreed hourly assumption and label it. Time capacity is not automatically cash saved: a team may use that capacity to respond faster or take on more work. Do not describe those possibilities as revenue already earned.
5. Make the next decision explicit
At the review, choose one of three next steps: continue the workflow, improve a specific weakness, or stop. Add another workflow only once the first shows enough value after supervision and operating costs. A useful pilot can also tell you what not to automate.
A practical exercise: bring one ordinary example, one exception and a rough weekly time estimate to your first conversation. Please remove client names and sensitive data.