Running a live quiz without turning the room into a contest
Projected quizzes motivate part of the group and expose the rest. How to get participation with no leaderboard, no names and no accounts.
The projected quiz has become a reflex. It wakes a room after lunch, and it produces something few exercises do: an answer from everybody at once.
It also carries a structural flaw that gets little attention: in its most common form, it ranks people against each other, in front of their colleagues.
The leaderboard problem
A score table on the wall does two things at once. It motivates the three or four people at the top, and it teaches everybody else that being wrong costs something publicly.
In corporate training the attendees are not strangers: they are colleagues, sometimes a manager and their team. The social cost of a wrong answer is not what it is in a room of anonymous students.
Somebody afraid of being seen to be wrong answers what they think is expected. That is exactly the data you do not want.
The mechanism is this: as soon as answering becomes risky, answers drift towards what looks correct rather than what the person believes. The exercise keeps producing numbers, but they stop measuring understanding.
What to project, and what not to
The useful distinction is not between showing and not showing. It is between aggregate and individual.
- Project: the spread of answers, the success rate, the shift between two questions. That makes the room discuss without naming anybody.
- Never project: a name, a ranking, an individual score, or anything from which you could reconstruct who answered what.
The trainer does need the per-person detail: that is their job, and it is what lets them follow up with somebody at the break. The difference is the audience, not the data.
Participation is lost before the first question
The second obstacle is logistical. Every step between "the trainer announces the quiz" and "the person answers" costs part of the group: download an app, create an account, recover a password, accept terms.
In a room of twenty, three of those steps lose a third. And the third you lose is not random: they are the least comfortable with the tooling, often the same people who would have benefited most from the question.
A link and a code read aloud are enough. It is the same reasoning as delayed evaluation, where friction destroys the sample.
Open one question at a time
A questionnaire opened all at once gets filled at different speeds, and part of the room reads ahead while the trainer is still talking. Releasing one step at a time keeps the group together and lets each result be discussed before moving on.
In practice
SkillLens works this way by default: the trainer opens steps one at a time, attendees join by link and code with no account, and what is projected is always aggregate, and no public view contains a name.