Understanding Similarity Scores

When GleanMark finds potential conflicts, each result includes information about how it was discovered and how similar it is to your mark. This guide explains the matching methods and how to interpret the results.

How Similarity Is Measured

GleanMark uses multiple matching methods to find potentially conflicting marks, mirroring how a USPTO examining attorney searches:

Phonetic Matching (Double Metaphone)

Finds marks that sound alike, even with different spelling. This catches the most common source of trademark confusion — consumers hearing a brand name and confusing it with another.

Examples: FONE ↔ PHONE, KOOL ↔ COOL, LITE ↔ LIGHT

Trigram (Visual Similarity)

Compares the character patterns of two marks to find ones that look alike on paper. Uses overlapping 3-character sequences to measure visual resemblance.

Examples: STRIPES ↔ STRIPS, GOOGLE ↔ GOGLE, NETFLIX ↔ NETFLEX

Semantic / Foreign Equivalents

Applies the Doctrine of Foreign Equivalents — if your mark contains a foreign word, GleanMark finds the English equivalent and vice versa. Also catches synonym relationships.

Examples: SOLEIL ↔ SUN, AQUA ↔ WATER, BLANC ↔ WHITE

Component Analysis

For multi-word marks, breaks the name into parts and checks if individual words appear in other marks, either alone or combined with other terms.

Examples: BLUE MOUNTAIN → checks BLUE, MOUNTAIN, and existing marks containing either word

Consonant Skeleton

Strips vowels to find marks with the same consonant structure, catching variations that preserve the "backbone" of a word.

Examples: STRIPE → STRP, STRIPS → STRPS

Discovery Method Badges

In Knockout Search and Clearance Report results, each conflict shows badges indicating which methods found it:

• Exact Match — direct text match
• Pseudo-Mark — match on USPTO's textual reading of a stylized mark (e.g., "1+1" → "ONE PLUS ONE")
• Design Code — design-element (figurative) code match
• Sounds similar — phonetic, sounds-alike match
• Looks similar — visual / character-pattern (trigram) match
• Common element — shares a word or root with your mark
• Foreign equivalent — match via the Doctrine of Foreign Equivalents
• Similar meaning — semantic / meaning-based match

A conflict found by multiple methods is stronger. If a mark triggers Sounds similar AND Looks similar AND Common element, it's a more significant risk than one found by a single method.

Risk Levels in Knockout Search

Knockout Search combines mark similarity with commercial overlap (how related the goods/services are) to produce a risk level:

Near-identical marks always score High regardless of class distance — an obvious look-alike is a risk even in a different product category.

Tips

• Phonetic matches are the most important — the USPTO weighs sound similarity heavily in Section 2(d) refusals
• Don't ignore medium-risk results — a mark with moderate phonetic similarity in the same class can still be cited
• Check the discovery method badges — multiple methods finding the same conflict = stronger risk
• Foreign equivalents matter — SOLEIL for a sunscreen brand will conflict with SUN-based marks in the same class

Related Features

• Knockout Search — See similarity scoring in action
• Running a Clearance Report — Comprehensive analysis with scored conflicts