To show a learning programme worked, you need three things a completion rate can't give you: a baseline reading of the people before you design it, a follow-up reading once they're back at work, and a claim about the difference that's bounded enough to survive a hard question. This guide is how to get all three.
Where measurement stops
Most programmes are measured at the applause. Attendance, satisfaction and completion arrive on time, fit the dashboard, and describe delivery, not change. Three months later, when the facilitator is gone and the pressure returns, no one asks what remained. The measurement system works as designed; it stops where the learning function stops watching.
This guide is the practical companion to the essay Measurement that cannot demonstrate return.
1 · Baseline — read the cohort before design closes
A baseline isn't a formality; it tells you what the intervention is about to land on. Before you finalise the design, answer five questions:
- How ready is this cohort to receive challenge — rested and open, or stretched thin?
- Where is resilience thin, and what will that cost under pressure?
- Which cognitive preferences dominate — and which learners will the chosen format disadvantage?
- What patterns are likely to surface when the work gets hard?
- What will you measure again, and when, after people return to work?
Measure readiness before the spend. It is the first evidence of return and the first design decision — not a number filed for later.
2 · Follow-up — measure again once the work resumes
One assessment is a photograph: accurate, but unable to show movement. Change becomes visible when you measure the same people again, at a chosen point, under stated conditions, and treat the distance between the two readings as the evidence. One score starts a conversation; two scores show a pattern. The industry has spent years improving the photograph while avoiding the cost of making the film.
3 · Bounded claim — say only what survives a hard question
A heroic claim (“the programme drove the results”) loses credibility the moment someone names another cause — a price change, a competitor exit, a new territory model. A bounded claim is harder to write and easier to trust:
- State the observed change.
- Name the comparison evidence.
- Write down the assumptions.
- List the plausible alternative explanations.
- Describe the contribution the programme likely made — and no more.
Measurement exists to make a claim that survives an intelligent question. That is what earns budget.
For finance — the hidden buyer
When the CFO asks “what changed?”, they are asking for a line between investment and movement. A bounded contribution claim gives them that, without pretending every human outcome reduces to money. Some of the most consequential learning — restored confidence, a reframed career — will never fit a dashboard. Say so. A function that claims measurement everywhere is trusted nowhere; let the demonstrated portion carry the weight, and let the human remainder stay human.
For how this runs inside an organisation — pilot, cohort diagnostics and leadership reporting — see LearnSmartly for organisations.
Read the people before you design, measure them again after the work, and make a claim you can defend.
FAQ
How do you prove a learning programme actually worked?
Measure the cohort before you design it, measure the same people again after they're back at work, and make a bounded claim about the difference — one that names its comparison evidence and its alternative explanations. Completion and satisfaction can't do this.
What is a learning baseline, and why measure it before the programme?
A baseline reads the cohort's readiness, resilience and processing preferences before design closes, so the programme is built for the people who'll receive it — and so you have a first point to measure change against.
Why measure learning more than once?
A single assessment is a snapshot; it can't show movement. Two readings, taken under stated conditions, turn a snapshot into evidence of change.
What is a bounded ROI claim?
A claim that states the observed change, names the comparison evidence, writes down its assumptions, lists alternative explanations, and describes only the contribution the programme likely made — so it survives scrutiny from finance.
See it on one of your own teams.