Method

How we count results

Automation agencies like to quote hours saved. This page states exactly how we measure ours, so you can check any number we give you, including the one in your own trial-run report.

Last updated:

In short

Every trial run starts by measuring the process as it runs today: cases per month and minutes per case, taken from your system logs where they exist, otherwise from a one-week tally by the people who do the work. “After” figures come from the system's own run logs over at least four weeks of live operation, not from a demo. Client figures appear on this site only with written client sign-off on the number and the wording.

Before we build: the baseline

  • Every trial run starts by measuring the process as it runs today: cases per month and minutes per case, taken from your system logs where they exist, otherwise from a one-week tally by the people who do the work.
  • Euro figures use the hourly cost you give us for the role doing the work, not an industry average.
  • The baseline is written into the trial-run report, so both sides agree on the starting point before anything is built.

After go-live: what counts as saved

  • Hours saved per month = (minutes per case before − minutes per case after) × cases per month ÷ 60. “After” includes review, corrections, rejected drafts and manual rework across all cases.
  • Estimated monthly value = hours saved per month × your hourly cost, minus monthly running costs: model usage, hosting, licences and ongoing care. Freed time is capacity, not automatically cash saved. Setup costs and payback are reported separately.
  • “After” figures come from the system's own run logs over at least four weeks of live operation, not from a demo.

What we leave out

  • Theoretical capacity: hours that could be freed if everything went perfectly.
  • Cases where a person discarded the agent's draft and did the work from scratch. Those count as manual.
  • Your team's one-off effort during setup, which we report separately instead of netting it away.

What we report next to the number

  • The override rate: how often a person changed or rejected the agent's draft.
  • Which steps stay with a person, and why.
  • The measurement period and how many cases it covers.

When a number is published

  • Client figures appear on this site only with written client sign-off on the number and the wording.
  • A figure not yet confirmed by run logs is labelled as an estimate.
Robert Van Ysendyck

Which task would you like to automate?

30 min · remote · no obligation. And if AI isn't the right tool, we'll say so.

Robert Van Ysendyck · Founder & Managing Director