What learning analytics means
Most online course tools record what learners do: who enrolled, which lessons were opened, how far videos were played, which quizzes were attempted and what scores resulted. Learning analytics is the practice of looking at those records to understand how a course is working. Used well, it points an instructor toward parts of a course worth a second look. Used carelessly, it produces confident-sounding conclusions about people that the data cannot support.
The data you will usually see
Different tools offer different dashboards, but a few measures appear almost everywhere.
- Enrollment and starts. How many people signed up and how many opened the first lesson.
- Completion. The share of learners who reached the end, or finished a set percentage of lessons.
- Progress by lesson. How many learners reached each step, which shows where people drop away.
- Quiz and assignment results. Scores, number of attempts, and which questions were missed most often.
- Time and activity. Time spent in lessons, video play position, login frequency.
- Learner questions. Comments or messages tied to specific lessons.
What this data can help with
Finding where learners drop away. Suppose, hypothetically, 100 learners start a course. If 80 open lesson 2, 74 open lesson 3 and then only 45 open lesson 4, the large drop between lessons 3 and 4 deserves attention. Losing 29 of 74 learners is about 39 percent, much larger than the steps before it (6 of 80, or 7.5 percent, and 20 of 100, or 20 percent). The numbers do not say why, but they tell you where to look: the lesson might be too long, too hard, unclear, or simply placed at a point where people get busy.
Spotting confusing quiz questions. If most learners miss the same question, including those who do well elsewhere, the question may be ambiguous or the lesson may not have covered the point clearly. Reviewing the wording is a sensible first step, and good online quizzes are written so that a miss reflects the learning, not the wording.
Checking whether objectives are being met. If you wrote clear learning objectives, you can see whether the related assessment results line up with them.
Prioritizing updates. Instead of rewriting everything, you can improve the few lessons where the data points to trouble, an approach that fits well with updating a course over time.
What the data cannot tell you
- Why someone left. A learner who stopped may have finished what they needed, changed jobs, run out of time or gotten stuck. Completion is only a proxy for success, and some learners with no intention of finishing still get what they came for.
- Whether they learned. Opening a lesson or playing a video is not understanding. A learner may let a video run in the background. Quiz scores measure what the quiz measured, which may be only a narrow part of the skill.
- Who is engaged. Time spent can mean deep focus, or confusion, or an open tab. Short time can mean a quick learner or a skipped lesson.
- Anything outside the platform. Notes taken on paper, discussions with colleagues and practice projects leave no trace in your dashboard.
Avoiding over-interpretation
Some habits keep conclusions modest.
- Mind small numbers. With 12 learners, one person leaving changes a percentage by more than eight points (1 in 12 is about 8.3 percent). Percentages from small groups swing wildly, so report the raw counts too.
- Compare like with like. A self-paced course taken by people with different goals will not have the same completion pattern as a cohort-based course with deadlines.
- Avoid single-number judgments. A low completion rate does not automatically mean a poor course, and a high one does not mean a great one. Look at several measures together, and read what learners say.
- Do not rank individuals on thin data. A dashboard that sorts learners by activity invites conclusions about people based on very little.
- Remember correlation. Learners who watch more videos may score higher, but motivation could explain both. Be careful about claiming that one thing causes another.
- Check your own tools. Differences in how a platform counts a “view” or “completion” can change numbers. Read the definitions before trusting a trend.
Privacy and fairness
Learner records are personal data. Collect and keep only what you need for improving the course, and think about who can see individual-level information. Tell learners plainly what is tracked and why, ideally in a privacy policy page written in everyday language. If you work with children or in an organization with its own rules, additional requirements may apply, and you should check them instead of assuming.
Where possible, prefer summaries about groups to monitoring individuals. If you do look at an individual’s data to offer help, treat it as a reason to reach out kindly, not as a verdict. Also consider fairness: learners with unreliable connections, shared devices or disabilities may show different activity patterns for reasons unrelated to effort.
Combining numbers with human feedback
The best use of analytics is as a starting point for questions. When a lesson stands out, ask learners directly, using a short survey or an open comment, and compare what they say with what the numbers show. Responding to those who have questions, as in supporting learners who have questions, often reveals the cause faster than any chart. When you present trends to colleagues or in a report, simple, honest charts built from the right chart types, with the number of learners clearly stated, are easier to trust than complicated dashboards.
A reasonable routine
A modest routine is enough: review progress by lesson and the most-missed quiz questions after each course run, note two or three things to examine, make small changes, and compare the next run. Keep notes on what you changed so that you can interpret later results fairly. Over time this builds a clearer picture than any single report.