Innovbon
How we measure results
A result is only worth publishing if you could check it yourself. These are the rules every number on this site follows.
We agree the metric before we start
Each project begins with one number we both want to move, such as review time per report or stock written off per month. Where possible we take a baseline from at least three months of data before any change.
We compare like with like
We measure against the same period last year or against a control group, so seasonal swings don't count as results.
The client signs off the figures
No result is published until the client has checked the numbers and approved them in writing.
We round down, never up
We show the measured value and the period it covers. If a figure falls between two numbers, we use the lower one.
We update or remove results that no longer hold
Every published result is reviewed at least once a year. If it no longer holds, we update it or take it down.
How strong is the evidence?
Every result carries one of these labels:
- Named client: the client is named and has approved the result.
- Client anonymised at their request: we describe the business instead of naming it, and a contact there will take a reference call.
- Pilot result: measured during a pilot or proof of concept, with the sample size and duration stated.
- Example scenario: an illustration, not a client result. These never appear in our client results.
Not sure where AI fits in your business?
Tell us about one problem that costs you time or money. We'll come back with a practical first step, usually within a week.