What is Similarity Score?

Search & Clearance4 min readUpdated Mar 25, 2026

A numerical value quantifying how closely a potentially conflicting trademark resembles the proposed mark across various dimensions.

A similarity score is a numerical value that quantifies how closely a potentially conflicting trademark resembles a proposed mark. Typically expressed as a percentage or a value on a defined scale, the score provides an objective, comparable metric that helps users prioritize potential conflicts by severity. A mark with a similarity score of 95% demands immediate attention; one scoring 40% may warrant a closer look but is unlikely to pose a serious obstacle.

Similarity scores are usually calculated across multiple dimensions, each reflecting a different way that marks can be similar. A phonetic similarity score measures how alike the marks sound. A visual similarity score measures how alike they look. A conceptual similarity score measures how alike their meanings are. Some systems also calculate an overall composite score that weights and combines these individual dimensions into a single number. The specific algorithms and weighting schemes vary by provider, but the goal is always the same: translate a complex, multi-dimensional judgment into a number that enables efficient triage.

It is important to understand that similarity scores are heuristic tools, not legal determinations. A high similarity score indicates that the marks share characteristics that would typically concern a trademark examiner, but it does not mean that the marks are legally confusingly similar: that determination depends on additional factors like the relatedness of the goods, the strength of the prior mark, and the conditions of purchase. Conversely, a low similarity score does not guarantee that no conflict exists, particularly if one dimension (such as conceptual similarity) is highly relevant but underweighted in the scoring algorithm.

Why It Matters

Similarity scores transform trademark clearance from a purely subjective exercise into a structured, data-driven process. Without scores, a clearance search that returns 200 results requires a human reviewer to examine each one and make a judgment call about its relevance, a time-consuming process prone to inconsistency and fatigue. With scores, the same 200 results can be instantly rank-ordered by risk, allowing the reviewer to focus their expertise on the highest-scoring matches and quickly dismiss low-scoring results.

For organizations that conduct clearance at scale (law firms handling dozens of searches per month, brand agencies screening hundreds of name candidates, or in-house IP teams managing large portfolios), similarity scores are essential for operational efficiency. They enable consistent triage across different searchers and different time periods, reducing the variability that comes from relying solely on individual judgment. They also make it possible to set threshold-based policies: for example, "all results scoring above 70% require attorney review" or "names with any result scoring above 90% are automatically eliminated."

How Signa Helps

Signa explains every search result. Each hit carries a match tier (identical, near_identical, similar or related) and the channels that found it, such as fuzzy, phonetic, embedded or lookalike, so a reviewer can see why a mark surfaced. Logo comparison is planned and not available yet.

Each hit also carries a relevance_score from 0 to 100. It orders the results and is not a likelihood of confusion: Signa does not calibrate it against conflict outcomes or map it to risk levels. Tiers describe how a mark was found, not a legal conclusion. Developers can use the tier and channels to sort results for review, for example reading identical and near_identical hits first, while the judgment about confusion stays with a qualified reviewer who weighs the goods and services, the strength of the prior mark and the market.

Example

A consumer goods company is evaluating an acquisition and wants a first view of the target's trademark portfolio as part of due diligence. Its IP team searches each of the target's marks through Signa's API in the classes where the target sells.

The team sorts the hits by match tier. Marks with identical or near_identical hits held by other owners in the same classes go to the top of the review list, marks with only similar or related hits come next, and marks with no close hits are noted as such. Attorneys then review the flagged marks and assess the actual risk, which feeds into the valuation discussion. The tiers organize the review; the conclusions are the attorneys'.