Most LMS reports are read the way a weather forecast is read: a glance, a nod, nothing changes. Completion sits somewhere in the sixties, the average score somewhere in the seventies, somebody screenshots both into the monthly deck, and the next batch goes through exactly the same course.
Reports only earn their keep when a specific number leads to a specific action by a named person within a week. That takes fewer reports than most teams run, read more often, with a short meeting attached.
This post sets out the five LMS reports worth reading every week, what each one tells you to do, how to spot a batch that is sliding before the exam confirms it, how to review questions properly, and the meeting rhythm that turns all of it into teaching decisions.
The five LMS reports worth reading weekly
1. Activity in the last seven days
Not completion. Just who opened anything at all. This is the earliest signal you have, and it moves days before scores do. Sort by cohort, and look for a cohort where activity has halved rather than for individuals.
2. Progress against plan, by cohort
Where each cohort should be in week four versus where it is. A single percentage across all learners hides everything. The useful version has one row per cohort with expected and actual side by side.
3. Stuck points inside a course
The lesson or module where learners stop. If forty learners reach module 5 and eleven finish it, module 5 is the problem, not the learners. This is the report that changes content rather than chasing people.
4. Assessment results by question
Not the average score. The pass rate per question, and how many attempts people needed. A single badly written question can pull a batch average down by several marks and teach you nothing about what they know.
5. Certificates issued and expiring
What was completed and, where certificates carry an expiry, who lapses in the next sixty days. In compliance-heavy work this is the report that keeps you out of trouble.
Report to action, not report to slide
| Report | What you look for | Action within the week | Who acts |
|---|---|---|---|
| 7-day activity | A cohort whose activity has dropped by half | Trainer calls the cohort lead and two learners, finds out why | Trainer |
| Progress vs plan | Any cohort more than one week behind | Add a live doubt-clearing session, or move the deadline once and say so | Programme owner |
| Stuck points | A module with a drop-off much larger than its neighbours | Open the module, cut it in two or add an example, republish | Course author |
| Question-level results | Questions almost everyone fails or almost everyone passes | Rewrite, retire or reteach the topic before the next batch | Subject expert |
| Certificates expiring | Anyone lapsing in 60 days | Re-assign the refresher and notify the supervisor | Administrator |
Spotting a struggling batch early
By the time a final assessment shows a weak batch, the course is over and you are writing a post-mortem. The signals that arrive early are behavioural, not academic.
- Login gap: the median days since last login for the cohort, not the average. One very active learner hides ten absent ones in an average.
- Pace: modules completed per week against the plan, tracked from week one rather than at the midpoint.
- First-attempt scores on early quizzes: a clear fall against the previous batch on the same quiz is worth a conversation now, not at the end.
- Time on module: both directions matter. Far below the expected time usually means clicking through; far above usually means the material is not landing.
- Forum and message volume: a cohort that has gone quiet is rarely a cohort that has understood everything.
- Repeat attempts on the same quiz: three attempts on one quiz by many learners points at the teaching, not at the learners.
Pick two of these, set a threshold in advance, and let the system flag the cohort. A threshold agreed beforehand removes the argument about whether the number is bad.
Question-level review
This is where an LMS earns back its cost, because it gives you data a classroom test never did. Two established measures do most of the work.
Difficulty index, or facility value, is the proportion of learners who answered the question correctly. Published guidance on item analysis puts the recommended range at roughly 30 to 70 per cent, with items above about 85 per cent treated as easy and those at or below 50 per cent as hard.
Discrimination index asks whether the stronger learners did better on this question than the weaker ones. On a scale to 1.0, a widely used classification treats 0.40 and above as good, 0.30 to 0.39 as reasonably good, 0.20 to 0.29 as marginal and below 0.20 as poor. Another commonly cited threshold treats 0.35 and above as good and below 0.20 as unacceptable.
The practical point is the combination. A question everybody gets right tells you nothing about who learned. A question with low discrimination is usually ambiguous, mis-keyed or testing something the course never taught.
Reading the distractors
For multiple-choice questions, look at which wrong options people chose. Item analysis guidance suggests retiring distractors that no one ever selects, and treating a distractor chosen as often as the correct answer by taught learners as a sign of ambiguity or a genuine misconception.
That second case is the gold. If half the batch picks the same wrong option, you have found a misconception worth fifteen minutes of teaching, and you found it before the exam. More on building question banks that do this in our note on an assessment engine that measures learning.
A simple review loop
- After each batch, export results for every question in the bank.
- Flag questions outside the difficulty range you have agreed and questions with low discrimination.
- Have the subject expert read each flagged question aloud. Most bad questions are obvious when read.
- Decide one of three: rewrite, retire, or keep and reteach the topic.
- Log the decision against the question so the next reviewer does not repeat the argument.
- Replace retired questions before the next batch so the bank does not shrink.
The meeting rhythm
Reports change teaching only when a meeting exists to act on them. Keep it small and fixed.
- Weekly, 20 minutes, trainers only. Three reports on screen: activity, progress against plan, stuck points. One action per cohort, with a name and a date. No slides.
- Monthly, 45 minutes, trainers and the programme owner. Question-level review of the last batch, content changes agreed, certificate expiries for the next quarter.
- Quarterly, 60 minutes, add the business owner. Batch-to-batch comparison on the same assessments, retirement of courses nobody takes, and whether the programme is changing anything at work.
Two rules keep these from decaying. Every meeting ends with actions written where everyone can see them, and every meeting starts by reading last week’s actions aloud. Skip the second and the first stops happening within a month.
Frequently asked questions
Is completion rate a useful measure at all?
It is useful as a floor, not as a goal. Completion tells you people got to the end. It says nothing about what they can now do. Pair it with first-attempt assessment scores and it becomes meaningful.
How many reports should a trainer look at?
Three every week and two every month is plenty. A dashboard with twenty tiles gets read once. The test is whether the trainer can say what she will do differently after reading them.
What is a good average score?
There is no universal figure, and chasing one distorts the questions. What matters more is the spread, the first-attempt rate and whether the same topics fail every time. A high batch average built on easy questions is worse than a lower one built on well-discriminating questions.
Our batches are small. Do these numbers still work?
Item statistics need reasonable numbers to be stable, so with a batch of fifteen treat them as a prompt to read the question rather than as a verdict. Pool results across several batches for the same question and the picture firms up.
Getting the reports in the first place
None of this works without per-question and per-cohort data you can export. Quipu LMS provides progress analytics, question banks across multiple-choice, short answer and descriptive formats, an assessment engine with auto grading, and certificates with expiry tracking, so the five reports above come out of the system you already run.
