How to find new restaurants worth trying

Last updated 3 September 2026

Finding restaurants is easy; every map app will hand you two hundred. Finding one worth a whole evening is a filtering problem, and most advice skips the filtering. This guide covers where good candidates actually come from, how to read the signals once you have some, and why the people you eat with are the best filter you own. It sits on Booth’s site; everything before the last section works without Booth.

Where candidates actually come from

Each source is good at one thing and lies to you in one way. Use two or three and let them disagree.

• The map, searched specifically. Not “restaurants near me” but the thing you actually want: “ramen”, “wine bar with food”, “somewhere with vegan mains”. A specific search surfaces the places that are actually about that thing; a generic one surfaces whoever ranks.
• City press and lists. Good for openings you would never walk past, and for neighbourhoods you do not know. Check the date: a “best of” list from three years ago is a museum catalogue, and half its entries have changed chefs or closed.
• People, asked the right question. “What’s good around here” gets you a shrug or a default. “Where did you eat recently that you’d go back to” gets you a real answer with a reason attached, which is the only recommendation worth having.
• Your own feet. A full room on a Tuesday, a short menu, a queue of locals: signals no platform can fake. Walking a neighbourhood before dinner beats scrolling one.
• Short video, treated as advertising. Food video is edited to make anywhere look transcendent, and the queue outside a viral spot is made of people who watched the same clip. Cross-check against the map before you commit an evening to it.

Reading the signals

• Count beats score, past a point. A 4.3 from thousands of people is an institution with a bad parking situation; a 4.9 from a dozen is a rounding error, or the owner’s family. Read the score and the crowd size together.
• Recent reviews outrank the average. The average remembers a chef who may have left. The last three months tell you what the kitchen is doing now.
• Look at photos of the food, not the room. A place that photographs its interiors is selling the interior.
• Price level is a fit question, not a quality question. The best night out is the one nobody quietly resented paying for. Settle the budget with the group before you fall for somewhere.

The filter is the people you eat with

Here is the part most guides miss: for a group dinner, “worth trying” is not a property of the restaurant. It is a property of the restaurant and your table. The tasting menu everyone raves about is worth nothing if it has one vegan option and your table has two vegans. So the last step of finding is proposing: put your two or three promising candidates to the people who are actually coming, and let their objections do the final cut. An objection with a reason (“I was there in March, it’s deafening”) teaches you something no star rating can. If dietary needs are part of your table’s picture, collect them before you shortlist, not after: here is what each restriction actually excludes.

Where Booth fits

Booth is a free web app for exactly that last step: a group decides where to go out in one room, with approvals, reasoned vetoes, and everyone’s dietary needs sitting beside every candidate. Its front page has the restaurant search, free and signed out, and a promising find pushes straight to your group from there. Whether it beats a poll or your group chat for your particular table is an honest question with an honest page: group dinner apps, compared.