Michelin stars vs. Tabelog: why neither is a bible

Two names decide where most people eat in Japan: the Michelin Guide and Tabelog. One is a century-old European institution with anonymous inspectors; the other is Japan’s own vast, crowd-sourced database. Both are genuinely useful. But the most important thing to understand about them is what they aren’t — and it’s the same for both: neither is a bible.

What Michelin sees, and what it misses

Michelin’s strengths are real. Its inspectors are anonymous and, in principle, consistent; it applies more or less the same yardstick around the world; and a star still means something. For a first-time visitor, it’s a reliable shortlist of serious kitchens.

But it comes with a particular lens. The guide grew out of French fine dining, and it rewards a certain register — refinement, polish, the trappings of a “destination” meal — better than it captures a brilliant eight-seat counter or a sticky neighborhood izakaya. Its coverage in Japan is also narrow: it inspects a handful of regions (Tokyo, Kyoto–Osaka and a few others) and simply doesn’t visit most of the country. A perfect rural soba shop can’t lose a star it was never in line for. And the standards, however rigorous, are opaque and slow to move.

What Tabelog sees, and what it misses

Tabelog is the opposite animal: Japan’s own diners, hundreds of thousands of restaurants, from three-star sushi to the ramen counter under the train tracks. It’s current, granular, and it covers the casual gems Michelin never reaches. For finding where locals actually eat, nothing beats it.

The catch is the number. Tabelog’s five-point scale is brutally compressed: as of 2022, only about 420 of its ~830,000 listed restaurants — roughly 0.05% — scored 4.0 or higher. A 3.5 is genuinely very good; a 3.8 is exceptional. That makes the scores useful but easy to misread, and it makes tiny gaps feel larger than they are. The algorithm behind them is opaque, too — opaque enough that a restaurant chain took Tabelog to court over how it weighted reviews and treated chains. Reviewer demographics tilt it as well. The crowd is wise, but it has its own moods and blind spots.

First: neither is a bible

Put plainly, both are filters, not verdicts. A starred restaurant can leave you cold: I’ve sat at a one-star Tokyo sushi counter that was flawless to look at — serene cypress, immaculate plating, faultless service — and frankly below average where it actually counts, on the rice and the fish. A 3.4, meanwhile, can be one of the best meals of your life; and some of the finest places in Japan sit in neither — no Michelin coverage, too few Tabelog reviews, or a chef who simply doesn’t court either. Treat a star or a score as a reason to look closer, never as the final word. The map is not the meal.

Second: ask the people around you

The best recommendation almost never comes from a list. It comes from a person who knows food and knows you. Ask a chef where he eats on his night off. Ask the regular two seats down. Ask a hotel concierge who genuinely eats out, a sommelier, a local friend whose taste you trust. A good human guide does the one thing no ranking can: they calibrate to your palate, and to this moment — what’s in season, what’s good right now, what suits the night you’re having. (That seasonal angle outweighs any score — see shun.)

Use Michelin and Tabelog to narrow the field. Then put the guidebook down and ask someone you trust to point you the rest of the way. (And if you’re just starting out, breadth beats prestige — see new to sushi.)