What this is
Four columns — L1 Reaction, L2 Learning, L3 Behavior, L4 Results — and three rows: what we’ll measure, how we’ll capture it, who owns the signal. It forces the evaluation conversation to happen at kickoff, not after launch.
Fill it in together in 45 minutes and you leave with a shared definition of success and a plan to capture it. If the L3 or L4 columns come out blank, that’s the finding: the training isn’t tied to a business outcome yet — better to learn that now than at launch.
Where xAPI changes the game
Traditional evaluation leans on completion records and a smile-sheet — self-reported, weeks late, silent on behavior. Instrument the experience with xAPI and the L3 column moves from “survey managers in six months” to “we can see it next week”:
- L2 → L3 in the data, not in a survey. A simulation that emits statements shows how someone passed — “knows the rule” separated from “applies it under pressure.”
- Attribution you can defend. Statements tie a person to a specific behavior at a specific moment, so the L3→L4 link stops being a leap of faith.
- A live signal, not a post-mortem. The same statements roll up to a manager view while there’s still time to coach.
Why I won’t start without it
It puts the uncomfortable question on the table early — can we actually attribute behavior change to this, or are we just hoping? — while there’s still time to design for the answer.
If you need the Kirkpatrick primer
The four levels, oldest framework in the book and still the shared vocabulary sponsors know: L1 is whether people liked it, L2 whether they learned it, L3 whether they do it on the job, L4 whether the business metric moved. Most programs report L1 and completion because those are easy. The canvas exists to force explicit choices about L3 and L4 — the two levels anyone funding the work actually cares about.
See it concretely: the Practice Room streams a conversation to an LRS turn by turn; the manager dashboard shows those statements once a leader has to act on them.