Our evaluation currently establishes that people came, that material reached them, and that they valued it. It does not yet establish that anything changed in the settings they went back to.
That is the gap, and this is what we are doing about it.
A post-event survey measures the room at the moment of maximum enthusiasm. Our 2025 figures came from 142 responses and were strong, but they measure perceived value on the day, from people who chose to attend and chose to respond. They cannot tell us what a manager did differently in October.
That limitation is printed in our impact framework. Printing it created the obligation to address it.
Follow-up measurement at six months, asking one question rather than twenty: name one thing you have changed. Open text, low burden, honest about non-response.
Organisational commitment tracking, with partners who agree in advance to name a specific change and to let us report whether it happened. Fewer participants, verifiable results.
Framework adoption as a distinct measure. Not downloads, which are exposure, but documented use: a policy that references the protocols, an induction that includes them, a process redesigned against them.
Lower numbers than our current reporting implies. A six-month behaviour question will not return 86 per cent. It will return a smaller figure attached to something real, and the difference between those two numbers is the honest cost of measuring properly.
We would rather publish the smaller number.
Because announcing a measurement after you know the result is not measurement. If we describe the instrument now, we cannot quietly discard it later if it is unflattering.
That is the point of publishing method: it commits you.