How AnyCheese Ranks Cheeses

AnyCheese rankings are designed to be data-driven, repeatable, and transparent.
They do not measure taste or quality. They measure fit, definition, and interest.

Each ranking follows the same mechanical system.

Step 1: Eligibility Filtering

Every ranking begins by defining which cheese types are eligible.

Eligibility rules may include:

  • Country of origin — for regional lists (e.g., French cheeses)
  • Texture or style — for category lists (e.g., soft cheeses)
  • Milk type — for milk-based lists (e.g., goat's milk cheeses)
  • Minimum data requirements — cheeses must have sufficient information

Only cheese types that pass all applicable filters are considered for scoring.

Step 2: Five Scoring Signals

Each eligible cheese type receives a score based on five independent signals:

A. Category Purity (0-3 points)

For texture-based rankings (e.g., "soft cheeses"), this measures how precisely a cheese matches the category.

  • Exact match: 3 points (e.g., "soft" matches "soft")
  • No match: 0 points (e.g., "semi-soft" does not match "soft")

This is an all-or-nothing score. Partial matches are not counted to ensure category focus.

B. Data Completeness (0-3 points)

Cheese types are rewarded for having comprehensive, well-defined information.

Five data categories are checked (maximum 3 points awarded):

  1. Country of origin or specific regional origin
  2. Milk type (cow, goat, sheep, etc.)
  3. Texture (soft, hard, semi-soft, etc.)
  4. Flavor or aroma description
  5. Age or rind information

Each populated category earns 1 point, capped at 3 total.

C. Popularity by Page Views (0-2 points)

Measures how often people view a cheese type's detail page.

Views are bucketed into three tiers using percentiles:

  • Top tier (above 67th percentile): 2 points
  • Middle tier (33rd-67th percentile): 1 point
  • Bottom tier (below 33rd percentile): 0 points

Percentiles are calculated within each ranking's eligible set.

D. Search Demand (0-2 points)

Measures how often users explicitly select a cheese type from autocomplete search results.

Only confirmed selections count:

  • User types in search box and clicks a suggested cheese type
  • Free-text searches without selection are excluded
  • Partial inputs are excluded

Like views, search selections are bucketed into three tiers:

  • Top tier (above 67th percentile): 2 points
  • Middle tier (33rd-67th percentile): 1 point
  • Bottom tier (below 33rd percentile): 0 points

E. Authority Bonus (0-1 point)

Cheese types with protected geographical status (PDO, PGI, etc.) receive a 1-point authority bonus.

This recognizes officially recognized cheese varieties.

Step 3: Weighted Scoring

Each of the five signals can have a configurable weight multiplier.

The final score is calculated as:

Total Score = (A × weight_purity) + (B × weight_completeness) + (C × weight_views) + (D × weight_search) + (E × weight_authority)

Different rankings may emphasize different signals, but the calculation method remains consistent.

Step 4: Sorting and Selection

Cheese types are sorted by total score in descending order.

Tiebreaker: If two cheese types have the same score, the one with more page views ranks higher.

The top N cheese types (typically 25) are selected for the final ranking list.

What the Rankings Are — and Aren't

They are:

  • Based on real user behavior and data completeness
  • Mechanically calculated using consistent rules
  • Updated automatically as data changes
  • Transparent and auditable

They are not:

  • Taste judgments or quality ratings
  • Expert opinions or editorial picks
  • Sponsored or paid placements
  • Static lists

Why This Approach

Cheese discovery has long relied on awards, reviews, and subjective lists.
AnyCheese takes a different approach: measuring how people actually interact with cheese information at scale.

The goal is not to decide what tastes best—but to surface what is most relevant, well-defined, and sought after.

By combining category precision, data quality, and real user interest signals, the rankings reflect both substance and demand.

Was this page helpful?