Learning analytics for BigBlueButton, at institution scale
The class ends. The picture stays.
You already ran the sessions. Every one of them is kept and made comparable, so a whole term becomes something you can look at. Who spoke and for how long, how much of the talking was the teacher, who sat through it without doing anything at all. Built and run by a BigBlueButton Certified Service Provider.
The analytics layer
Dashboards, not downloads.
Filter by date, by meeting or by the person, and everything on the screen answers for what you picked. It is the same view your team opens next month, and the month after that.
Filters
- Date
- Meeting
- Participant
Meetings
12,480
Participants
38,214
Total meeting time
9,412 h
Average meetings a day
138
Average duration
47 min
Public chat messages
214,880
Users / Meeting per Day
Sessions run, against the people in them
SessionsParticipants
Features usage
Share of sessions that opened each tool
Meeting duration by time of day
Teaching load, weekday against hour
User roles
Who was in the room, and as what
Viewer 82%
Moderator 15%
Presenter 3%
Polls Response Rate
Of everyone who could have answered
Meetings Metrics
One row per meeting, over the range you picked
| Meeting | Sessions | Learners | Spoke | Whiteboard |
|---|---|---|---|---|
| BIO-201 | 164 | 38 | 71% | 24% |
| LAW-118 | 132 | 42 | 54% | 61% |
| ENG-330 | 96 | 29 | 83% | 12% |
| MTH-105 | 210 | 51 | 38% | 48% |
Meeting name word usage
What staff actually called their sessions
Daily Whiteboard Moves
Drawing activity across the year
Monthly Meetings
One square a day, across the year
Darker is busier. The gaps are the vacations.
What is already built
11
Dashboards, grouped into five families
46
Charts across them, ready on the first day
12
Kinds of chart, from a single figure to a heatmap
100%
Of your sessions kept, course after course, term after term
Built to your question
The twelfth dashboard is the one only you ask for.
Eleven dashboards cover what every institution asks. The one that matters most to you is usually not among them, and this is the part a fixed panel can never do. If you can describe the question, we can build the view.
- “Do our evening sessions run differently from the daytime ones?”
- “Which rooms stopped using the whiteboard after week four, and did their chat go up?”
Step 01
You ask in your own words
No query language, no ticket template. A sentence from a programme director in an email is enough to start, and if the question turns out to be two questions we will say so.
Step 02
We model it against what is already there
Almost every new question is a new way of looking at what we already keep, rather than something new to collect. That is why the answer is usually a date rather than a project.
Step 03
It appears beside the others
Same filters, same permissions, same exports as everything else. It is not a one-off report that gets emailed round once. It is a dashboard your team opens again next term without asking us.
What the platform answers
The questions institutional analytics asks, answered inside the classroom.
An institutional analytics platform sees a virtual class as one attendance mark, because that is all the LMS hands it. BBB Analytics starts where that mark ends, inside the session, where the talking and the chat and the polls and the whiteboard are.
Who took part, and who did not?
Attendance is not participation. The difference is visible at a glance.
AttendedSpoke at least once
Was this a lecture or a conversation?
The balance between the person leading and everyone else.
ModeratorViewerPresenter
What did the session actually use?
Slides, whiteboard, polls, chat and breakout rooms. How many, and how often.
How is the course going, not just this class?
The same questions asked across a term instead of a morning.
Who opens it
Five people, five questions, one place to answer them.
A university decides as a committee. A dean, a programme director, the person teaching the seminar, the quality office, the team running the servers. Five people who will never be in the same meeting, each arriving with a different question. These are the ones we are asked.
Dean or pro-vice-chancellor
“Across a year of teaching, which sessions ran as a discussion and which as a broadcast?”
The synchronous didactic ratio, computed on every meeting: the share of the talking that was the moderator’s. One session is an anecdote. A year of them is an argument.
Programme or course director
“Which groups went quiet as the term went on, and in whose sessions?”
Non-participation rate and session silence ratio, meeting after meeting, with an engagement timeline inside each one. Direction is visible while there is still time to act on it.
Lecturer or seminar leader
“Of the thirty-eight in the room, how many did anything at all?”
Every participant carries an academic score and the longest they went without acting. The ones who attend everything and are never heard are invisible in a register and plain here. It is a prompt to reach out, not a mark.
Quality and accreditation
“Show me the evidence behind the engagement claim in the annual monitoring report.”
Reviews ask for evidence about teaching that nobody kept. This keeps it, at the level a report is written at, and exports it in the format a panel expects.
Head of digital education, or IT
“Which hours are we sized for, and did the tools we trained staff on in September get used by November?”
Load by hour of day and day of week, and tool diversity tracked meeting by meeting rather than surveyed. Both come out of the same tables.
How deep it goes
From the number to the reason.
Most reporting stops at the figure. The question that changes anything is the next one: why does it look like that. Answering it means going from a year of meetings down to one participant in one session without changing tool.
The number
What happened. Counts, durations, shares and distributions, for the live teaching that until now left no record behind it.
The reason
Why it looks like that. Go from a year of meetings down to one participant in one session. Cut the same measure by hour of day, by role, by what launched the meeting, or by breakout room against the main one.
Data governance
Where the data lives, and who can open it.
For a university this is not one criterion among five. It is the gate. So it is answered here rather than in a later conversation, and what follows is how the platform is actually run.
- One named place
- The platform runs on RIADVICE infrastructure, in one location that is named in your agreement rather than left vague. Nothing is spread across a shared analytics cloud you cannot point at.
- Opened by role
- A quality lead and an external reviewer each open the dashboards you granted them and nothing else. Access is given in the platform, not by forwarding a spreadsheet.
- A platform of your own
- If your policy says the analytics cannot sit anywhere shared, we will stand up a dedicated platform for your institution alone. Tell us that early and we scope it with you.
- What it reads, and what it does not
- It reads the events BigBlueButton already produces about a session, including the chat your teachers can already see. No recording is analysed, no camera is read, and nothing tries to infer attention or mood from any of it.
- GDPR, answered by the architecture
- A university has to say where personal data sits, who processes it and under what agreement. The platform runs in one named place under a contract with us, so those questions have short answers. We put them in writing before a pilot.
A note on method
What these figures are, and are not.
Everything here describes what can be observed in a recorded session: who spoke, who wrote, which tools were opened. These are not measures of attention, motivation, understanding or learning outcomes. They exist to support a professional conversation about teaching, not to replace one and not to judge a person.
BigBlueButton Certified Service Provider
Published and supported by RIADVICE
The team behind BBB Analytics is a BigBlueButton Certified Service Provider and hosts BigBlueButton for institutions every working day. The figures and the people who can explain them come from the same place. A question about a number does not become a support ticket.
Point it at a term you have already run.
Bring one course. We will model it, stand the dashboards up on your data, and show you the estate-level view before you commit to anything.