How we rank businesses
A composite score, weighted, with the math published. No pay-to-play. No surprise tier bumps. Sponsored placements (when they exist) are clearly labeled and never affect ranking.
The dominant signal in the composite. Built from review aggregation across multiple sources, with community-submitted reviews getting 25% of the within-Quality weight, external aggregations getting the remainder. Every aggregation is Bayesian-smoothed with a category-mean prior so a place with three glowing reviews cannot leapfrog a place with three hundred mostly-glowing ones. Thin-data businesses get pulled toward the category average until enough signal accumulates to justify a higher (or lower) rank.
Internal signals from the people using this site. Saves, votes, clicks, return visits. A pick that locals keep coming back to and recommending climbs. A pick that no one ever bookmarks fades, even if its quality score looks fine. Confidence math caps low-data businesses so brand-new entries cannot dominate.
Three small signals that share the bucket equally. Editor: a local with taste actually shows up and writes notes (transparent on each business page). Trust: chamber/merchant-association affiliation, BBB membership, claim status (verified owners get a small bump). Freshness: last-verified date and review velocity. A pick that has not been touched in two years gets stale and loses points until someone re-verifies.
Buzz across neighborhood social channels, AI-classified for theme and tone. We never quote, screenshot, or republish anyone's posts. We extract paraphrased themes ("locals love the patio," "service has slipped lately") and weight them. Small share by design: social is the noisiest signal and we treat it as a tiebreaker, not a driver.
Why Bayesian smoothing matters
Without smoothing, a coffee shop with a single five-star review beats a coffee shop with four-hundred reviews averaging 4.6. That is obviously wrong. The category-mean prior pulls every score toward what is typical for the category until enough reviews accumulate to overcome the prior. Confidence math caps how high a low-data place can rank, no matter how good its average looks.
What we never do
- Quote, screenshot, or republish text from review platforms or social channels.
- Fabricate quotes or reviews. Paraphrased themes only.
- Sell ranking position. Sponsored placements are labeled and ranked separately.
- Hide score changes. Every adjustment is logged.
Last updated: 2026-05-24