Every year, L&D teams send the same number up the chain: completion. Ninety-eight percent finished the course. The report goes green, the box gets ticked, and everyone moves on.
The problem is that completion measures one thing: that people reached the last screen. It says nothing about whether they remember what was on the screens before it, and even less about whether they work any differently on Monday.
What completion actually measures
Completion is an attendance record. In a self-paced course it often records something smaller: that someone clicked Next enough times. Most learning platforms cannot tell the difference between a person who read every word and a person who left the course running in another tab.
That is why completion is so easy to push to 98%. Send three reminders, copy the manager, and the number climbs. None of those reminders teach anything.
A green completion report can sit right next to a red audit report. They are measuring different things.
Why the number survives
If completion says so little, why does every dashboard lead with it? Because it is easy. Every learning platform reports it by default, it fits in one cell of a spreadsheet, and it only ever goes up when you push harder.
Regulators play a part too. For mandatory training, an auditor often asks one question: did everyone take the course? Completion answers that question perfectly. The mistake is letting the auditor's question become the business's only question.
There is also a quieter reason. A completion number cannot fail you. A memory check or a business metric can. Leading with completion protects the training team, but it leaves the business guessing whether the money worked.
People forget fast, and courses are built as if they don't
Researchers have known for more than a century that memory fades quickly without practice. New information that is not used or recalled fades within days. A single 90-minute session, taken once a year, is close to the worst possible design for remembering anything.
The fix is not more content. It is less content, used more often, in situations that look like the job: a decision at the counter, a call from an angry customer, a claim that does not add up.
Three things to measure instead
1. Decisions, weeks later
Give people a short, realistic scenario 30 days after the course. Can they still make the right call? This is the closest you can get to measuring memory without watching people work.
2. The work metric the course was built to move
Every course should exist because something in the business is going wrong: repeat audit findings, support tickets, errors, early exits. Agree on that number before you build anything, and report it after launch.
3. Time to competence
For onboarding and new systems, measure how long it takes someone to do the job on their own. It is a number managers already care about, and it moves when training is good.
What this means for your next course
Start from the metric, not the content. Write down the one number the course should change, then design backwards: which decisions move that number, and what practice makes those decisions automatic?
Then plan the 30-day check before launch day, not after. If the course works, you will have proof your CFO understands. If it does not, you will know before next year's renewal.
- Pick the one business number your next course should move.
- Cut every screen that does not help a learner make a decision.
- Schedule a short scenario check 30 days after launch.
- Report completion if you have to, but lead with the business number.
Founder of Yuto Learning. Placeholder bio: one or two lines on his years building training for global companies and why he started Yuto.
