Measuring completion honestly
Completion rate is the number most often quoted and most easily arranged. The denominator is where the arranging happens.
Course marketing quotes completion rates the way maintenance quotes planned-versus-reactive: as a headline number with a definition nobody states. Reported figures vary by an order of magnitude between courses that are, in substance, similar.
Most of that variation is definitional rather than real.
For a workplace example of turning activity records into trend data rather than a single headline metric, workforce analytics software covers analytics at team level.
Decide the denominator and say it
Completion of what, among whom? Everyone who paid, including those who never logged in? Everyone who started the first module? Everyone who was still active at the midpoint?
Each is defensible and they produce very different numbers. The most informative single figure is completion among learners who completed the first module, because it separates the course's performance from the aspirational-purchase effect. Whatever you choose, state it, and use the same one over time.
For general operational guidance relevant to small organisations, the U.S. Small Business Administration provides public business resources.
Start rate — how many who paid began. Completion rate — how many who began finished. Collapsing them into a single number hides which of the two problems you have.
Define finished
Watching every video, submitting every exercise and passing the final assessment are three different definitions, and platforms usually default to the weakest.
For a course sold on capability, the figure worth tracking is the one tied to the outcome: how many produced the artefact or passed the assessment. That number is lower and it is the one that means something.
Look at the shape, not the total
The aggregate tells you how many finished. The per-module curve tells you where they stopped, which is the only version that supports a fix.
A cliff at one module points at that module. A steady decline points at a length or scheduling problem. A drop right after the first assessment points at the assessment. The total conceals all three.
Track elapsed time as well as completion
A learner who finishes in nineteen weeks a course designed for six has completed it, and something is nonetheless wrong. Median elapsed weeks is a leading indicator that the effort estimate is off, and it moves before the completion rate does.
Be careful what you optimise
Completion is easy to raise by making the course shorter and the assessment easier. Both improve the metric and can degrade the product.
The number worth watching alongside it is whether learners can do the thing afterwards — a follow-up a few months later, or evidence of what they built. Harder to collect, and it is the only check that keeps the completion figure honest.
Publish it if you quote it
If completion rates appear in sales material, publish the definition next to the number. Courses that do this stand out immediately, and the transparency is more persuasive than a higher figure with no definition.