Polls that change the session, not decorate it

A poll is only worth running if the answer alters what happens next.

Most polls online are theatre. They are inserted to prove interactivity, the results appear, everyone looks at the bar chart for a moment, and the session continues exactly as it would have without them. Audiences learn this quickly and stop answering, which is the correct response to being asked a question that has no consequence. The remedy is not fewer polls. It is polls with teeth.

A poll earns its place when the running order genuinely branches on the result. Ask how familiar people are with the subject, and then actually skip the introduction if the answer says to. Ask which of three problems matters most, and then spend the middle of the session on the winner, having prepared all three. Ask at the end what remains unclear, and let that set the agenda for the next instalment. The cost of this is preparation: you must be ready to deliver more than you will use. The benefit is a session shaped by the people in it.

There is a diagnostic use as well. A poll answered by a small fraction of the audience is telling you something about attention, and a poll whose answers cluster far from what the speaker assumed is telling you the session is aimed wrongly. Both signals are more valuable than the poll's nominal subject. Read them honestly, adjust while there is still time to adjust, and note them for the design of the next one.