How to decide where to eat without the twenty-minute conversation

"Where do you want to eat?" is one of the most reliably unproductive questions in the language. Not because nobody has a preference — most people do — but because the question is open-ended, everyone is being polite, and the cost of naming a place is looking like the person who made everyone eat Thai.

The fix is not better manners. It is a better question.

Why the conversation stalls

Three things happen at once.

Everyone defers. "I don't mind" is rarely true. It means "I have a mild preference and I am not willing to spend social capital on it."

The option space is infinite. Asking someone to name a restaurant is asking them to search their memory of every restaurant they know, filter by what is open, what everyone eats, and how far people are willing to walk. That is real work, done under mild time pressure, for no reward.

Nobody wants to be responsible. If you choose and it is mediocre, that is on you. Deferring is the safe move, and if everyone plays it safe the conversation cannot end.

Notice that none of these is solved by more discussion. More discussion makes all three worse.

Method 1: two options, one vote

Narrow it to two and put it to a vote. This is the fastest method and it works because it removes both the search cost and the blame.

• Nobody has to think of a restaurant, only choose between two.
• Nobody is responsible for the outcome, because the group chose.
• It takes under a minute.

The key move is that one person picks the two candidates. That feels undemocratic and is precisely why it works — it converts the hard, open-ended part into a small, low-stakes decision that one person can make quickly, and leaves the group the part that is easy and fun.

If you want the vote counted properly rather than by scrolling back through reactions, make a two-option poll and drop the link in the chat. People can vote without an account, so nobody has to install anything.

Method 2: the veto round

Everyone gets one veto and no votes. Someone proposes a place; anyone can kill it once. In practice this ends in two or three rounds, and it surfaces the constraints people would not otherwise mention — allergies, "I had that yesterday", "that place is forty minutes away".

Good for groups where the real problem is unspoken constraints rather than indifference.

Method 3: rotating dictator

One person decides, and next time it is someone else. No discussion at all.

This sounds worse than it is. Over a few outings everyone gets their turn, the average outcome is about the same as a group decision, and you get back the twenty minutes. It works best for groups that eat together often.

Method 4: constrain first, choose second

Do not start with restaurants. Start with the constraint that eliminates the most options:

1. How far are we willing to go?
2. What are we not eating tonight?
3. Sit-down or quick?

Answer those three and the list is usually down to two or three places on its own. Then vote.

Method 5: the default

Have a place you always go when nobody decides. Not the best restaurant you know — a reliable one that nobody objects to. The point of a default is that it ends the conversation instantly when nobody cares enough to override it, which is most of the time.

Which to use

The general shape

This is a specific case of a general pattern: a group decision stalls when the question is open-ended, and unblocks when someone converts it into a small number of concrete options. That applies to restaurants, to which film to watch, to where to go on holiday, and to most of what teams argue about at work.

The work is in the narrowing, not the voting. Whoever is willing to do the narrowing is doing the group a favour, even if their two options are imperfect — because two imperfect options that end in a decision beat infinite perfect ones that end in "I don't mind."

If it helps, this guide covers how to phrase the two options so the vote is worth counting, and the this-or-that list has a food section you can steal from.