Methodology

How Matcha Crawl picks are organized.

The current board blends directory completeness, matcha-specific quality signals, and community ratings as they grow. We do not sell placement, and we avoid movement labels until there is enough real usage data to support them.

1

You rate a spot

On any shop page, slide the matcha meter from 0 (never again) to 10 (life-changing). That’s it — one number, no sign-up required.

2

We keep it fair

One vote per device per day, per shop. We use Cloudflare Turnstile to block bots — no CAPTCHAs for you, no ballot-stuffing for them.

3

Math does the rest

We average all votes for a shop, then pull that average toward the citywide mean (7.5) until enough people weigh in. A shop with one perfect 10 won’t leapfrog a spot with fifty solid 9s.

4

The board updates

Shops need at least 3 votes to appear on the leaderboard. Below that, they show as “Not yet ranked” — we’d rather wait for signal than guess.

The nerdy bit (optional reading)

For community ratings, we use Bayesian averaging — the same formula review sites use to prevent a single vote from dominating. In plain terms: every shop starts near the city average, and ratings gradually pull it toward its observed score as more people weigh in. The formula weights toward the global average of 7.5 until a shop accumulates enough ratings for its own average to carry more of the result.

When vote counts are shown, they help you judge how established a community rating is. Directory pick scores may also reflect completeness, source freshness, matcha focus, and editorial quality markers.