What an LMS Actually Records from a SCORM Course
8 min read Czytaj po polsku
An LMS knows surprisingly little about what happens inside a SCORM course. It knows what the course tells it through the SCORM API, nothing more. Once you know that short list, most "why does the report say…" questions answer themselves. Field names are given for both SCORM 1.2 and 2004; if you haven't picked a version yet, start with SCORM 1.2 vs SCORM 2004.
The full list of what an LMS receives
An LMS receives eight kinds of data from a SCORM course: completion, pass/fail, score, session time, a bookmark, suspend data, the exit mode and interactions.
| What | SCORM 1.2 | SCORM 2004 | Example |
|---|---|---|---|
| Completion | cmi.core.lesson_status |
cmi.completion_status |
completed |
| Pass/fail | cmi.core.lesson_status |
cmi.success_status |
passed |
| Score | cmi.core.score.raw |
cmi.score.scaled (+ raw) |
85 / 0.85 |
| Session time | cmi.core.session_time |
cmi.session_time |
0000:14:32 / PT14M32S |
| Bookmark | cmi.core.lesson_location |
cmi.location |
slide-9 |
| Saved state | cmi.suspend_data |
cmi.suspend_data |
course-specific string |
| Exit mode | cmi.core.exit |
cmi.exit |
suspend |
| Answers | cmi.interactions.n.* |
cmi.interactions.n.* |
one record per answer |
Data also flows the other way. The LMS gives the course the learner's ID and name, whether this is a fresh start or a resume (cmi.core.entry or cmi.entry), and the pass mark from the manifest (cmi.student_data.mastery_score or cmi.scaled_passing_score).
Everything else is invisible to the LMS: clicks, time per slide, how much of a video was watched, which tabs were opened. A course can pack such things into suspend data, but the LMS can't report on them.
Completion: the course decides
The course decides when it is complete and tells the LMS. The LMS doesn't watch the slides. Typical rules:
- All slides visited. Simple, but it rewards clicking Next. A 12-slide course can be "completed" in 90 seconds.
- Required activities done, for example a branching scenario played to an ending.
- Quiz passed.
- A combination, such as all slides visited and quiz passed.
If completion has to mean something, pair slide visits with a required activity or a quiz.
SCORM 2004 adds another option. The course sends cmi.progress_measure from 0 to 1, and the LMS compares it with a completion threshold from the manifest: with 0.8, the learner is complete at 80% progress.
In SCORM 1.2, cmi.core.lesson_status holds one value. When the course sends passed after completed, "completed" is gone. Most LMSs count passed as complete, but check the setting.
Pass/fail and score
Pass/fail and a score exist only when the course has a graded assessment. In SCORM 1.2 the course sends cmi.core.score.raw (usually 0–100) and sets status to passed or failed. In 2004 it sends cmi.score.scaled from -1 to 1, usually with the raw score, and sets cmi.success_status.
If the manifest also defines a passing score, the LMS can recalculate the result from the score. Keep the course's pass mark and the manifest's value identical, or learners will see "Passed" on screen and "Failed" in the report.
A course without a graded quiz has nothing to score and reports completion only. That's how Scormbee works too: without a graded quiz, the course reports completion with no pass mark.
Time: an open window, not attention
The course reports how long each session lasted, and the LMS adds sessions into cmi.core.total_time or cmi.total_time. It measures how long the window was open. A learner who leaves the tab open overnight gets credited with hours. Don't use time alone as proof of anything.
Bookmark and suspend data
The bookmark stores where the learner stopped. Suspend data stores everything else the course needs to restore itself: answers, visited slides, scenario choices.
- Bookmark: 255 characters in 1.2, 1,000 in 2004.
- Suspend data: 4,096 characters in 1.2, 64,000 in 2004 4th edition. These are the minimums an LMS must support.
- The course chooses the format, often compact JSON. The LMS stores it as an opaque string.
- Both survive only if the course set the exit mode to
suspend.
Interactions: one record per answer
An interaction records one answered question: what was asked, what the learner answered, what was correct, the result, and how long it took.
| Field | SCORM 1.2 | SCORM 2004 | Example |
|---|---|---|---|
| ID | id |
id |
q3-sender |
| Type | type |
type |
choice |
| Learner's answer | student_response |
learner_response |
b |
| Correct answer | correct_responses.0.pattern |
correct_responses.0.pattern |
c |
| Result | result |
result |
wrong / incorrect |
| Latency | latency |
latency |
0000:00:14 / PT14S |
| When | time |
timestamp |
|
| Question text | not available | description |
"Which address is suspicious?" |
All fields sit under cmi.interactions.n., where n is the record's index. Types are true-false, choice, fill-in, matching, sequencing, likert, numeric and performance. SCORM 2004 adds long-fill-in and other. Results are correct, wrong (1.2) or incorrect (2004), unanticipated and neutral.
Response formats differ. SCORM 2004 separates choices with [,], so a two-option answer is a[,]c, and matching pairs look like 1[.]a[,]2[.]b. SCORM 1.2 uses plain commas and dots.
In SCORM 1.2 interactions are write-only: the LMS stores them but the course can't read them back. Each new answer usually gets a new index, so a learner who retries a question three times can appear three times.
Scormbee sends every checked answer as an interaction: quiz questions, true or false, matching, fill in the blanks, ordering, sorting, and the ending a learner reaches in a branching scenario.
How to read an interaction report
Read by question first, then by learner.
- Many wrong answers on one question: the content doesn't teach it, or the question is ambiguous. Reread the question before rewriting the slide.
- One wrong option chosen by many learners: a shared misconception. That deserves its own slide.
- Very short latency on correct answers: the question may be too easy, or the answer guessable.
- Very long latency: confusing wording, or learners looking things up.
Report screens vary a lot. Some LMSs list every record per learner, some aggregate by question, some show only the last attempt. If yours can export to CSV, a pivot table answers the questions above in minutes.
Why a course "stays incomplete"
Usually the course never reached LMSFinish (1.2) or Terminate (2004), the last save was lost, or the completion rule wasn't met. Check in this order:
- The window closed before the final save. Chromium-based browsers block synchronous requests while a page is closing, so a course that saves only at the end can lose its last data. Courses should call
LMSCommitorCommitafter every meaningful event. - A popup blocker. Many LMSs open courses in a new window. Blocked, the course never starts. Allow popups for the LMS domain or open the course in the same window.
- The completion rule wasn't met. A skipped slide, an unfinished activity, a failed quiz. In 1.2,
failedreplacescompleted. - Suspend data overflow. Past 4,096 characters in 1.2, the LMS may reject or cut the value, and later saves break.
- Exit not set to
suspend. Each launch starts a new attempt, and the report shows the newest one as incomplete. - API not found. Opened outside the LMS, through a direct link or a downloaded ZIP, the course records nothing.
- Expired LMS session. After a long break, saves fail with a login error.
To diagnose, reproduce the problem in SCORM Cloud (it offers a free trial account) and read the API log. Look for the finish call, errors on SetValue, and the last status sent.
Privacy: what is personal in SCORM data
The learner's identity comes from the LMS, and everything the course sends is stored against it, so treat it as personal data.
- The course reads the name and ID from
cmi.core.student_nameandcmi.core.student_id(1.2) orcmi.learner_nameandcmi.learner_id(2004). SCORM 1.2 suggests "Last, First" order, which matters if you print names on certificates. - A course never needs its own form asking for name or email. That would move personal data outside the LMS's controls.
- Fill-in answers are stored as typed. Don't ask open questions that invite personal details.
- Under GDPR, scores and answers linked to a learner are personal data. Retention is set in the LMS.
- Embedded YouTube videos and externally hosted files make the learner's browser contact third-party servers.
- A website ZIP without an LMS tracks nothing at all.
If you are building your first course, How to build a SCORM course in the browser covers the testing steps in detail.
Frequently asked questions
Does SCORM record which answers a learner chose?
Only if the course writes interactions. Each record can hold the question ID, type, learner response, correct response, result and latency. Whether you can see them depends on the LMS's report screens.
Why does my SCORM course stay incomplete?
Usually the course never called LMSFinish (SCORM 1.2) or Terminate (SCORM 2004), the last save was lost when the window closed, the completion rule wasn't met, or suspend data went over the limit. Reproduce it in SCORM Cloud and read the API log.
Is the time in an LMS report accurate?
It is the time the course window was open, summed over sessions. A learner who left the tab open over lunch gets credited for lunch, so treat time as a rough signal, not proof of learning.
Where does the learner's name in a SCORM course come from?
From the LMS. The course reads cmi.core.student_name (SCORM 1.2) or cmi.learner_name (SCORM 2004), so it never needs to ask the learner for it.
Does a course without a quiz report a score?
No. Without a graded assessment there is nothing to score, so the course reports completion only. In Scormbee, a course without a graded quiz reports completion with no pass mark.