How to write a poll people actually finish

Almost every abandoned poll is abandoned for one of three reasons: it asked too much, it asked ambiguously, or it never explained why it was asking. None of those are fixed by design tweaks — they are fixed by editing the questions.

Cut until it hurts

The strongest predictor of completion is length. A poll of five questions is not "slightly better" than one of fifteen; it is a different thing entirely, because the respondent decides whether to start based on how long the page looks.

Go through your draft and delete every question you cannot name a decision for. If you can't say what you would do differently depending on the answer, the question is curiosity, not research.

• One question per idea. "How fast and how helpful was support?" is two questions wearing one coat.
• No question that your own analytics already answers.
• No demographics you will not actually segment by.

Ask about behaviour, not intention

People are unreliable narrators of their own future. "Would you use this feature?" reliably overstates demand, because agreeing is free and polite.

Ask what already happened instead. "When did you last do X?" produces a far more honest signal than "Do you plan to do X?", and it takes the same amount of space.

Put the boring questions last

Age, role and company size feel like natural openers, but they are the least engaging thing you could lead with. Open with the question the respondent actually has an opinion about — by the time they reach the demographics they have already invested effort and will finish.

Say what happens next

A single line above the first question — what the results are for, and whether answers are anonymous — measurably lifts completion. It costs nothing and it is the difference between a form and a conversation.

If your poll is anonymous, say so plainly. If it is not, say that too; people are far more forgiving of attribution they were warned about than of attribution they discover afterwards.