How we work.

Every restaurant we cover earns an Insider Score: a single number from 0 to 10, calibrated to this city, rebuilt every week, and never for sale. It is not a raw average of other sites' stars. It is our own trust-weighted, momentum-adjusted, editor-confirmed read of the full review record. It weighs three things.

The inputs are the aggregate review record: ratings, review counts, hours, and operating status from Yelp, Google, and TripAdvisor, each used under its licensed terms. This is data journalism. We read what thousands of diners report over time, not one critic's single meal. No signal we use in a live ranking is more than 48 hours old.

A machine layer finds and scores candidates. A human layer confirms them and writes the editorial. Nothing on this site publishes without both passes.

Trust-weighted rating

The rating is the main input, weighted by how much evidence stands behind it. A five-star room with thirty reviews gets pulled toward the city's average; a 4.9 with three thousand reviews holds its ground. It is the same Bayesian weighting IMDb uses for its Top 250: trust a rating in proportion to the sample size it is built on.

Momentum

Among rooms with similarly trusted ratings, the one with more sustained recent traffic edges ahead. It is a tiebreaker, not the main driver. A 4.5 with twelve thousand reviews cannot leapfrog a 4.9 with three thousand, but it can hold its own in the top ten.

The result is a list of restaurants that have both the rating to back it up and enough sample size to prove the rating is real, not a fluke.

Insider Scores are calibrated per city. The highest-ranked room in Washington sets the top of the scale at 9.9, and everything below scales to that anchor. The three factors and their weights are identical across every city in the network; only the local anchor moves.

We do this because every city's dining scene has its own ceiling. Forcing one global scale would either flatten smaller markets into the sixes and sevens or inflate the biggest ones until every city looked the same. Per-city calibration tells the truer story: this is the best room in this town, ranked against the rooms it actually competes with. A 9.9 in Washington is the best Washington has. It is not a claim about any other city.

Data ages. Restaurants close, change chefs, move rooms. If you spot an error, stale hours, a closed room, a factual mistake in a guide, email the desk and we will fix it. Substantive corrections are logged in the open: the Updated date on the guide changes, and the entry is revised. We do not silently edit published work.