Every L&D dashboard includes a completion rate. Most have several — by cohort, by module, by quarter. It's the easiest number to produce, the easiest to report upward, and the easiest for a CHRO to nod at in a review meeting.
It's also almost entirely disconnected from the question that actually matters: is this changing what people do?
That disconnect isn't a data problem. It's a measurement design problem — and it's worth understanding precisely, because fixing it changes what you build, not just what you report.
Completion answers the wrong question by design
Completion rate was never built to measure capability. It was built to measure delivery — did the content reach the intended audience, did people engage with it to the end? That's a legitimate operational question for a training function to answer. The trouble starts when it quietly becomes the primary measure of success, because it's the number that's easiest to collect at scale.
Once completion becomes the headline metric, the incentive structure follows it. Content gets designed to be finishable — clear, digestible, low-friction — rather than designed to build a skill that's genuinely difficult to acquire. The two goals aren't the same, and they sometimes actively conflict: the training that's easiest to complete is rarely the training that does the most to change behaviour, because real capability-building involves friction, failure and repetition — the things a "finish this course" metric is structurally biased against.
What a better mental model actually measures
The useful reframe is to stop asking "did they finish it" and start asking a stage-by-stage question instead: at which point in the learning loop is this actually breaking down?
There's a simple four-part way to think about it. Did the learner leave able to describe what good looks like? Did they get a chance to actually attempt the real decision, not just read about it? Did they receive feedback on that attempt while it was still fresh, rather than in a generic end-of-course score? And did they get to repeat the attempt enough times for the right response to stop requiring conscious effort?
Most programmes are strong on the first of those four and largely silent on the other three. That's diagnostic information a completion rate can never surface, because it's not designed to look inside the loop — only at whether someone walked through the whole thing once.
The version of this that actually changes the CHRO conversation
If you want to shift how leadership thinks about L&D ROI, the individual number that tends to land hardest isn't a completion percentage — it's a transfer estimate, stated in your own currency. Take your annual spend, apply a conservative published transfer rate, and show the rupee figure that represents investment that never reaches behaviour. For most organisations, that number is larger and more uncomfortable than anyone expects, and it reframes the conversation from "did people attend" to "where exactly is this breaking, and what would it cost to fix that specific stage."
That's a harder conversation to have than reporting a completion percentage. It's also the one that actually predicts whether next year's training budget will produce a different outcome than this year's.
If you replaced completion rate with a single better number tomorrow — one that actually predicted behaviour change six months out — what would it need to measure?