Scoring
How the number works.
Every check-in produces a WingScore. Every WingScore contributes to a restaurant's aggregate. Here's the math — no mystery, no algorithm you can't read.
The WingScore
WingScore starts with a weighted rubric across five criteria. The weights are fixed and opinionated — they reflect what actually makes wings good or bad:
| Criterion | Weight | Why |
|---|---|---|
| Sauce | 29% | Flavor quality and execution — the most diagnostic criterion. |
| Crispiness | 29% | The skin. Tied with sauce; bad crispiness ruins good sauce. |
| Meat | 22% | Density, moisture, and pull. Less variable than the outside, but it counts. |
| Blue Cheese | 10% | Optional. If you order it and it's wrong, it matters. |
| Celery | 10% | Optional. A small signal, but a committed one. |
Blue cheese and celery are optional. If you mark either N/A, their weights (10% each) are redistributed proportionally across the remaining criteria. A bone-in basket without celery re-normalizes to sauce 32.2%, crispiness 32.2%, meat 24.4%, blue cheese 11.1%.
The Hot Take gets a say
Hot Take is the gut call you tap before the rubric. When it disagrees with the five criteria, it nudges the WingScore up or down. The nudge is capped, so instinct can sharpen the result without overruling the evidence.
A core failure still matters. If sauce, crispiness, or meat falls apart, the WingScore takes a standards penalty. Great sauce cannot hide dry chicken.
Read the 90-second versionThe restaurant aggregate
One person's WingScore doesn't make a restaurant's ranking. CAFO uses a Bayesian average to produce the aggregate shown on the map and leaderboard.
aggregate = (n × avg + m × C) / (n + m)
n = number of check-ins for this restaurant
avg= restaurant's raw average WingScore
m = minimum check-ins threshold (5)
C = global mean across all restaurants (7.2)
In plain English: a new restaurant starts near 7.2 and earns its way to its true average as more people check in. This prevents one person from putting a place at the top of the map by logging a 10/10 on day one.
Example
A spot with 2 check-ins averaging 9.5 gets: (2 × 9.5 + 5 × 7.2) / (2 + 5) = 7.7
A spot with 40 check-ins averaging 8.8 gets: (40 × 8.8 + 5 × 7.2) / (40 + 5) = 8.6
The established spot ranks higher despite a lower raw average — because 40 opinions mean more than 2.
The heat color
Score pins and badge colors follow a five-stop gradient interpolated across the 1–10 range:
Color is strictly a visual aid — it maps the same score to the same color everywhere in the app, including map pins, score badges, and breakdown bars.
Quick answers
- How is WingScore calculated?
- WingScore combines fixed weights for sauce, crispiness, meat, blue cheese, and celery. A capped Hot Take adjustment can nudge the rubric, and a standards-floor penalty can pull it back.
- What happens when blue cheese or celery is skipped?
- Skipped optional criteria do not become made-up scores. Their weight is redistributed proportionally across the criteria that were scored.
- Why can a proven spot outrank a higher raw average?
- Restaurant aggregates use a Bayesian average. New spots begin near the global baseline and move toward their raw average as more check-ins make the result more trustworthy.