What this note helps with

Use this note to choose stronger evidence without turning every learning project into an oversized analytics project.

What this feels like in the work

This feels like having numbers but still not having an answer. We can show completions, attendance, and maybe a survey score, but the leader is asking whether the work changed anything that matters.

The problem

Completions answer whether someone passed through the system. They do not answer whether the person is ready, whether a manager sees behavior change, or whether the workflow improved.

The system move

Choose measurement based on program stakes. Low-stakes content can use lightweight signals. Skills-based work needs application evidence. Priority programs need behavior and business-adjacent indicators designed before launch.

What to inspect

Look at whether the team has defined the decision that the evidence will support before the learning asset is built.

Use this when
  • The team can report completions but cannot explain whether the work helped
  • Leaders are asking for impact, but the program was not designed around evidence
  • Measurement feels like a post-launch reporting task instead of a design decision
Diagnostic questions
  • What decision will this evidence help a leader make?
  • What observable behavior would change if the learning worked?
  • What signal is credible enough for the stakes of this program?
What to watch for
  • A dashboard that answers participation but not readiness
  • A survey question that asks whether people liked the training but not what changed
  • A measurement plan that starts after the asset is already built
What this can look like
  • Completion as proof We can show participation, but not readiness, decision-making, task performance, transfer, or workflow impact.
  • Evaluation after launch The program is already built before anyone defines what evidence would make it credible.
  • Same measurement for every program A low-risk reference page and a high-stakes capability program get the same survey or completion report.
What we can try first
  • Match measurement depth to program stakes
  • Define readiness, behavior, adoption, and workflow signals before launch
  • Use completions as one signal, not the whole proof story
Small experiment

Before the next build starts, write one sentence that names the decision the evidence should help us make. Then choose one behavior signal that would make that decision less fuzzy.

Three ways we can respond
  • No technology Write one evidence decision statement: this program should help us decide whether to continue, change, scale, or stop something.
  • Microsoft 365 or Google Workspace Use Forms, Excel, Sheets, or PowerPoint to collect one manager observation, one learner confidence check, and one behavior example after launch.
  • AI-assisted Ask AI to draft scenario questions, manager observation prompts, behavior indicators, and a simple evidence summary from survey comments and work samples.
Sources