Measuring what changed.

Most training reports count attendance and satisfaction because those are the two things that can be measured without the client's cooperation. Neither tells anyone whether the organisation is better at anything.

A presenter points to bar and line charts on a screen while two colleagues listen in a lounge.

01The four levels

We agree the evaluation level before the programme, not after.

01ReactionDid participants find it relevant and well delivered. Collected every time, and the weakest evidence available.

02LearningDid knowledge and skill change. Measured through assessment, examination or pre- and post-programme testing.

03BehaviourIs the work being done differently. Measured through line manager observation and structured follow-up at an agreed interval.

04ResultsDid the operational number move. Only possible where the client supplies the data and accepts that other factors also moved it.

Where a return on investment figure is required, we apply the Phillips model on top of level four, with the assumptions stated on the page rather than buried in an appendix.

Two things we will not do.

Claim a result the data cannot carry

If turnover fell in the year we ran a leadership programme, the programme may have contributed. It did not cause it on its own, and a report that says otherwise is worth nothing the first time an executive asks a hard question about it.

Measure at level four without the client's data

Level four requires operational figures we do not own. Where a client cannot or will not release them, we say the evaluation stops at level three rather than constructing a number.