How to run a live quiz, start to finish
· 4 min read
This walks through one live quiz end to end: the decisions you make while building it, what you do in the room, and the twenty minutes afterwards that decide whether any of it mattered. It assumes you have run a meeting before and have never run a quiz.
Before you build anything, decide what the quiz is for
Three different jobs hide behind the phrase "run a quiz", and they want different decks.
A warm-up exists to get people talking to each other. Five questions, low stakes, speed-weighted scoring, a visible leaderboard. Accuracy barely matters.
A knowledge check exists to tell you what landed. Eight to twelve questions, generous timers, and you care about which questions the room got wrong, not who won.
An assessment exists to produce a record. Timers long enough that nobody is guessing, scoring that ignores speed, and results you will look at afterwards.
Building a warm-up when you needed a knowledge check is the most common way a quiz disappoints. The room enjoys it and you learn nothing.
Building the deck
Start with the questions that matter and cut backwards. Most first decks are too long. Twelve scored questions is roughly twenty minutes with discussion, which is longer than it sounds when the room is quiet.
Write distractors that a reasonable person would pick. A four-option question where three options are obviously wrong is a one-option question wearing a costume. If you cannot think of a third plausible wrong answer, use two options and move on.
Mix in something that is not a quiz question. A poll or a word cloud every four or five questions changes the rhythm and gives people who are behind on points a reason to stay engaged. They also produce the moments worth talking about, because there is no right answer to hide behind.
Put the hardest question about two thirds through. Opening with your hardest question suppresses participation for the rest of the session. Ending with it leaves the room on a low note.
Setting timers
Match the timer to the work, not to a house style.
| Question type | Starting point |
|---|---|
| Recall of a single fact | 10 seconds |
| Multiple choice needing comparison | 20 to 30 seconds |
| Calculation or reading a chart | 45 to 60 seconds |
| Anything you want discussed first | no timer, close it manually |
The live answered-count tells you when you got it wrong. If it hits full with half the timer left, the remaining seconds cost you attention. If it stalls at sixty percent, the question is harder or less clear than you assumed.
Choosing scoring
Speed-weighted scoring, where a faster correct answer scores more, produces energy and a leaderboard that moves. It also rewards the person who guesses immediately and punishes the person who thinks. That trade suits a warm-up and actively damages an assessment.
For a knowledge check, lengthen the timers so that speed stops discriminating, or turn timing off entirely. The room reads the difference within two questions and adjusts how carefully it answers.
Streak bonuses amplify whatever the base scoring already rewards. They make a runaway leader more likely, which is fun in a social quiz and demoralising in a classroom.
Running the room
Get everyone in before you start. Show the join code, wait for the count to stop climbing, then wait ten more seconds. Latecomers who join mid-quiz start from zero and usually disengage.
Say what the rules are. Whether it is anonymous, whether speed counts, whether there is a prize. Thirty seconds of this prevents the entire session being played on wrong assumptions.
Read each question aloud. People who are slow to look up, or whose phone is loading, get a second entry point. It also sets the pace, because the room starts answering when you stop talking.
Do not narrate the leaderboard every time. Showing standings after every question flattens them; by question six nobody looks. Three times in a twenty-question quiz, with something said out loud each time, makes each one an event.
Talk about the wrong answers. The reveal chart is the most useful artefact in the session. When sixty percent picked the same wrong option, that is a finding about your material, and saying so out loud is worth more than announcing the correct answer.
When something goes wrong
A phone will not join. Decide in advance: that person pairs with a neighbour. Say it at the start and it stops being an interruption.
The room is silent. Usually the question is ambiguous rather than hard. Re-read it, offer one clarifying sentence, and extend the time. If it happens twice in a row, the material is ahead of the room.
Someone is gaming the scoring. Speed-answering and early guessing are responses to incentives you set. If it is spoiling the session, lengthen the timers rather than asking people to stop.
Afterwards, within a day
The results have a short half-life. Everyone saw the charts, one person photographed one, and by Thursday nothing survives.
Send one screen: what you asked, what came back, and what happens next. Lead with the split rather than the average, because "eleven of you said four or five, seven said one or two" is a sentence people react to and "average 3.4" is not.
Name the two questions the room got worst and re-teach them. In a work setting, the equivalent is the process question everyone answered differently, which usually means the process exists in four versions.
Then convert one finding into one named action with one owner. Reporting results without a change teaches people that answering is theatre, and the next session's participation drops.
A reasonable first deck
If you want a shape to copy: ten scored questions, timers between fifteen and forty seconds depending on the question, one poll at question four and a word cloud at question eight, standings shown after questions five and ten, speed weighting on if this is social and off if it is not. Run it, note which questions stalled, and cut two before you run it again.