How many voters do you need?

It depends on what you want from the poll. A handful of people you trust can tell you whether one option is clearly loved or clearly ignored. That’s an opinion check. To tell apart two options that people like about equally, you need far more votes, and even then a poll measures what people say they prefer, not what they’ll do.

What a few votes tell you

Five or ten colleagues, friends or customers are enough to catch the obvious. If nobody puts an option in their top list, you can drop it. If almost everyone puts the same option first, you have a strong hint. A small poll also shows you what you missed: the logo you thought was safe that nobody picked, or the thumbnail you’d almost cut that keeps winning.

What a small poll can’t do is separate two options that are close. A 6 to 4 split between two finalists says very little.

When a result is too close to call

Here’s a way to see why. Say it comes down to two finalists, and each voter picks one. Suppose people had no real preference and picked by tossing a coin. How often would either option get at least this many votes, just by chance?

Voters Votes for the leading option How often coin tosses give either option at least this many
10 6 About 75% of the time
20 12 About 50% of the time
40 24 About 27% of the time

Each row is a 60% share, and each is easy to get by chance. To bring the chance below 5%, the leader needs 9 of 10 votes, 15 of 20 or 27 of 40. So with ten voters, only a lopsided result means much. With forty, a smaller lead starts to count.

These figures are plain coin-toss arithmetic (the binomial distribution), not a survey of real polls. Real voters aren’t coins, but the point holds: a narrow lead from a small group can easily be luck.

What researchers suggest

Two user research firms publish rules of thumb for formal studies. Nielsen Norman Group recommends 40 participants for most quantitative usability studies. Lyssna’s preference testing page says you can typically get quantitative results from 30 to 50 participants. For a quick poll among your own people, treat them as the point where “a hint” starts to become “a result”.

Who votes matters more than how many

Forty colleagues aren’t forty customers. People who already know your work, share your taste or want to be kind will vote differently from the people you’re trying to reach. A small poll of the right people beats a big poll of the wrong ones. If you can, send the link to people who look like your viewers, buyers or users, and keep your team in a separate poll.

Reading a Pairpoll result

Pairpoll gives you two views. The overall ranking adds up points from every finished voter’s top list. Each voter’s own list shows how much people agree.

  • If the same option sits at the top of most voters’ lists, the result is clear.
  • If the top places change from voter to voter, the options are close. Treat the leaders as a tie and decide on other grounds, or ask more people.
  • Only voters who finished count. Someone who stopped halfway doesn’t move the ranking.
  • Each voter’s pairs come in a shuffled order, so an option that met a strong rival early for one voter is likely to meet different rivals for the next. That evens out as more people vote. How voting in pairs works explains why.

A practical rule

Decide before you look what result would change your mind. If you only need to drop the weak options, a small group of the right people is enough. If you need to choose between two close favourites, either get dozens of the right people or accept that it’s a judgment call and make it.

All guides

Make your own poll

Free. Upload 2 to 16 images or videos and share one link. Voters don't sign up.

Create a poll

Pairpoll uses Google Analytics cookies to see which pages people visit, only if you accept. Details are on the privacy page.