Why a poll is the wrong tool for a small group

Every group chat has done it. Somebody drops a poll — Italian, Thai, or burgers? — three people tap Thai, two tap Italian, one taps burgers, and Thai wins. Then the person who tapped burgers mentions they can't eat anything with peanuts, the two Italian voters go quiet, and twenty minutes later you are back where you started.

The poll did exactly what it was built to do. It counted votes and named the largest pile. The problem is that counting votes answers a question a small group rarely asks.

What a poll measures

A vote is one bit of information per person per option: this one. It cannot say how much you wanted it, whether your second choice was a close second or a distant one, or whether the option you didn't pick is fine or unbearable. A three-two-one split looks decisive, but it hides everything that matters: the three Thai voters might have been mildly in favour, and the one burger voter might have had a hard reason.

Polls are a tool for large populations, where you cannot ask everyone how they feel and the majority is the best approximation available. In a group of six you can ask everyone how they feel. Using a poll there throws away most of the information you could have had.

What a small group actually wants

Ask people what a good outcome looks like for a dinner, a film or a weekend, and almost nobody says "the thing most of us voted for". They say some version of:

• Nobody is stuck with something they genuinely can't do.
• Nobody feels steamrolled.
• The thing we pick is good enough for everyone, even if it's nobody's first choice.

That is a compromise, not a majority. And a compromise has a different shape: it needs to know each person's strength of feeling about each option, it needs to treat "I can't" as absolute rather than as one vote, and it needs to be able to explain itself, because a group will only accept an outcome it understands.

Three things a poll can't do

It can't see intensity. Five people who would each be fine with option A and one person who loves option B looks like five-to-one for A. But if those five would also be fine with B, B makes everyone happy and A makes one person miserable. A poll cannot tell the two situations apart. A tool that asks everyone to rate each option — even a quick one-to-five — can.

It can't respect a hard no. An allergy, a budget ceiling, a night you are not free: these are not preferences to be outvoted. A majority cannot make a restaurant safe for someone with a nut allergy. The moment a group has one of these, the right answer is to remove that option entirely and choose among what is left. A poll has no way to say "this one is off the table". It just counts.

It can't explain the result. "Thai won, 3–2–1" is a fact, not a reason. When a group can see why — everyone was happy with the cuisine, it suited the kind of evening everyone had in mind, one person would have preferred something closer — the outcome feels fair even to the people who didn't get their first choice. A number alone never does that.

What to do instead

Ask everyone the same few questions, privately, and let something add them up honestly:

1. Rate every option, not just pick one. A one-to-five scale is enough to separate "love it" from "fine" from "please no".
2. Give everyone a veto, and make it absolute. One "I can't" removes an option for the whole group, and the group should see that it did and why.
3. Answer alone, before seeing anyone else's answers. The first strong opinion in a chat drags everything after it.
4. Show the reasoning, not just the winner. Who is delighted, who is compromising, and what was ruled out.

That is what Compromise does. You make a room, share a link, and everyone answers on their own. The result is the option the whole group can live with, with a plain-language explanation — "everyone is happy with this one", "two of five wanted something cheaper", "Anna can't accept this one" — so nobody has to take the answer on faith.

It takes about two minutes and needs no accounts. The next time somebody reaches for a poll, try a room instead.