Book a demo

Technical Feature Comparison: A Patentability Search Method

Patentability Search · Feature Analysis

A practical method for decomposing an invention into stable comparison units and recording what one prior-art reference actually discloses.

Technical feature comparison begins after potentially relevant prior art has been retrieved. The reviewer decomposes the proposed invention into clear comparison units, maps each unit to the prior-art disclosure, and records whether the reference provides explicit, implicit, or no disclosure.

Granularity matters. Units that are too broad hide meaningful distinctions; units that are too fragmented can separate features that operate as one claimed combination. This guide preserves a claim-aligned decomposition method, a feature matrix, novelty and inventive-step review, and three common-error corrections. It uses the EPO framework as the main legal reference; other jurisdictions may apply different tests.

What technical feature comparison and decomposition mean

Technical feature decomposition breaks the proposed technical solution into units that can be described, searched, and compared. The units should reflect the matter likely to appear in a claim rather than becoming an arbitrary list of words.

  • Descriptive: the unit can be expressed clearly in technical language.
  • Searchable: terminology, functions, relationships, or classifications can represent it in a search.
  • Comparable: the reviewer can map it to a specific passage, figure, or necessary implication in a reference.

The EPO Guidelines describe essential claim features as those needed to achieve the technical effect underlying the solution to the technical problem, and state that an independent claim must include the features presented as necessary to carry out the invention.1 That legal framework makes claim structure a useful starting point, although decomposition remains an analytical aid rather than a substitute for claim construction.

Where assisted comparison fits. Eureka IP Search’s Novelty Search extracts core technical features, builds multiple search strategies, and compares close prior art feature by feature.2 The user and qualified professionals still need to validate the feature model, reference content, relevant date, and legal conclusion.

The granularity problem: how fine should features be?

There is no universal feature count. The useful level depends on the proposed claim, the technical relationships, and the question being tested.

A practical standard: align with the proposed claims

When draft claims exist, start with the limitations of the proposed independent claim and then map dependent-claim additions separately. If claims do not yet exist, create a provisional claim-like statement that preserves the invention’s essential structures, materials, steps, functions, and relationships. Revise that model only for a documented technical or drafting reason, not to avoid a newly found reference.

Illustrative heat-management example

Assume an illustrative invention describes a mobile-device heat-management assembly with a graphene thermal film contacting a processor and a phase-change layer that absorbs heat.

Too coarse: “a heat-management device for a mobile phone.” This unit does not capture the proposed material, location, or cooperating structure.

Moderate granularity:

  1. a heat-management device includes a graphene thermal film;
  2. the film contacts a surface of the processor;
  3. the device includes a phase-change material layer; and
  4. the phase-change layer absorbs heat generated by the processor.

Too fragmented: treating “graphene” and “film” as unrelated units when the proposed limitation is the combined concept “graphene thermal film.” Fragmentation can obscure the claimed relationship even if each word appears somewhere in the art.

Two auxiliary questions

  • Does the concept operate as one claim limitation? Preserve a unitary relationship when splitting it would change the technical meaning.
  • Does the feature contribute to solving the stated problem? If removing it changes the proposed solution or effect, treat it as a core candidate and ask the drafter to confirm its claim status.

Operational method for technical feature decomposition

Step 1: identify problem, means, and effect

  • Technical problem: what technical issue does the invention address?
  • Technical means: which structures, materials, relationships, conditions, or steps provide the solution?
  • Technical effect: what result is linked to those means?

The technical means usually supply the comparison units. The problem and effect help explain why a unit matters and how inventive step may later be framed. However, do not turn an asserted effect into a proven fact without supporting disclosure or evidence.

At this intake stage, Eureka IP Drafting’s Invention Disclosure workflow can turn rough engineering notes into a structured disclosure and highlight missing details or overlooked inventive points.3 This is a separate pre-drafting workflow; it does not determine whether a prior-art document legally discloses a feature.

Step 2: distinguish core, additional, and contextual features

Working labelMeaning in the analysisComparison treatment
Core claim candidateA feature proposed as necessary to the independent-claim solutionCompare it against every reference used for the independent-claim novelty analysis
Additional or dependent featureA narrower limitation proposed for a dependent claim or fallbackRecord separately and compare when assessing that narrower claim set
Contextual or background elementInformation that explains the field but may not limit the claimDo not label it conventional or common knowledge without evidence

These are working labels, not legal conclusions. Whether a feature is essential, limiting, optional, or supported depends on the application and the claim language under review.

Step 3: construct a feature comparison matrix

IDTechnical featureWorking typeLinked effect
F1Graphene thermal filmCore candidateConducts heat away from the processor
F2The film contacts the processor surfaceCore candidateProvides a heat-transfer path
F3Phase-change material layerCore candidateAbsorbs heat during a phase transition
F4A specified film-thickness rangeAdditional featureBalances thermal behavior and package space

This is an invented teaching example, not a technical recommendation or a real claim. In a live matter, use the disclosure and proposed claims to define both the feature wording and the stated effect.

Methodology for technical feature comparison

Principle 1: compare feature by feature

Do not rely on an overall impression that a document is “similar.” For each feature, cite the precise paragraph, claim, figure, table, or other disclosure. Also record the relationship among features; finding separate components is not necessarily the same as finding their claimed arrangement.

Principle 2: apply implicit disclosure narrowly

A feature need not appear word for word. Under the EPO novelty test, a document can disclose features that a skilled person would consider implicit, but the claimed subject matter must be derivable directly and unambiguously from the document.4 Well-known equivalents that the document does not disclose belong to obviousness analysis, not novelty.

For example, silence about thickness does not automatically disclose every practical thickness. Mark a thickness limitation as implicit only when the disclosed teaching necessarily produces or requires the claimed limitation. If an extra selection, measurement, or modification is needed, record the gap instead of filling it with hindsight.

Principle 3: distinguish generic and specific disclosure

EPO guidance states that a generic disclosure usually does not remove the novelty of a specific example within it, while a specific disclosure can remove the novelty of a generic claim that includes it.5 Thus, “carbon material” does not automatically disclose graphene, while a disclosed graphene film may fall within a broader claim to a carbon-material film. Selection inventions and ranges can require additional analysis.

Comparison record format

FeatureD1 disclosureRecordReason
F1: graphene thermal film“a heat-dissipation layer made of graphene”ExplicitThe material and layer are stated
F2: contacts processor surface“attached to a CPU surface”ExplicitThe structural relationship is stated
F3: phase-change layerNo phase-change material identifiedNot foundD1 instead describes a metal heat sink
F4: specified thickness rangeA value inside the proposed rangePotential matchVerify units, measurement context, and range law

“Potential match” is a review flag, not a novelty verdict. The final legal assessment must consider the claim wording, the entire reference, direct and unambiguous disclosure, enablement, dates, and governing law.

Overall solution assessment versus individual features

Novelty: assess one disclosure as a whole

For EPC novelty, a reviewer does not combine separate prior-art items to assemble the claim. The EPO Guidelines also caution against combining separate embodiments within one document unless the document specifically suggests that combination.6 Therefore, if D1 discloses F1 and F2 but not F3, D1 does not anticipate a claim requiring F1, F2, and F3 merely because F3 appears in D2.

The test is stricter than checking whether every feature exists somewhere in the technical field. One reference must disclose the claimed combination, explicitly or implicitly, under the applicable novelty standard.

Inventive step: analyze the differences and the reason to combine

The EPO problem-solution approach identifies the closest prior art, establishes the objective technical problem, and asks whether the claimed solution would have been obvious.7 A feature matrix helps identify what D1 lacks and where D2 may be relevant, but the presence of F3 in D2 does not by itself prove obviousness.

The “could-would” inquiry asks whether the prior art would have prompted the skilled person to modify the closest prior art to reach the claimed solution, rather than merely whether the person could have combined the pieces.8 Record the proposed motivation, technical compatibility, expected result, and any hindsight concern separately from the novelty matrix.

Common technical feature comparison errors

Error 1: substituting an effect for a limiting feature

Incorrect: “D1 dissipates less heat, so the invention is novel.”

Correction: compare the claimed structures, materials, relationships, steps, and any properly limiting functional language. A better result may inform inventive-step reasoning or evidence, but an unsupported performance statement does not create novelty by itself.

Error 2: reading the reference first, then redefining the invention

Incorrect: after seeing D1, rewrite the feature model so the invention appears different.

Correction: establish and version the feature model before comparison. If the disclosure or proposed claim later changes, record who changed it, why, and which comparison must be rerun.

Error 3: treating functional similarity as automatic novelty disclosure

Incorrect: D1 says “heat-dissipation layer” and the invention says “thermal film,” so the terms are either automatically different or automatically equivalent.

Correction: examine the structure, material, function, context, and claimed relationship. For novelty, the question is whether D1 directly and unambiguously discloses the claimed feature, not whether two labels sound similar or serve a broadly related purpose.4

Key takeaway. Technical feature comparison works best when the units track the proposed claims, preserve meaningful combinations, and remain stable throughout review. Compare one reference feature by feature, apply implicit disclosure narrowly, distinguish generic from specific teaching, and keep novelty findings separate from inventive-step combination reasoning.

Sources and verification

  1. EPO Guidelines 2026, F-IV 4.5.2 — Definition of essential features. Verified July 2026.
  2. Patsnap Eureka IP Search. Verified July 2026.
  3. Patsnap Eureka IP Drafting. Verified July 2026.
  4. EPO Guidelines 2026, G-VI 2 — Implicit features or well-known equivalents. Verified July 2026.
  5. EPO Guidelines 2026, G-VI 4 — Generic disclosure and specific examples. Verified July 2026.
  6. EPO Guidelines 2026, G-VI 1 — Novelty and state of the art. Verified July 2026.
  7. EPO Guidelines 2026, G-VII 5 — Problem-solution approach. Verified July 2026.
  8. EPO Guidelines 2026, G-VII 5.3 — Could-would approach. Verified July 2026.

This article provides general search and comparison methodology, not legal advice. Claim construction, novelty, inventive step, enablement, relevant dates, and evidentiary rules vary by jurisdiction and matter. Qualified professionals should review the analysis.

Build a reviewable feature comparison

Structure the technical features, test multiple search paths, and compare close prior art with source-linked evidence and human review.

Explore Eureka IP Search

Your Agentic AI Partner
for Smarter Innovation

Patsnap fuses the world’s largest proprietary innovation dataset with cutting-edge AI to
supercharge R&D, IP strategy, materials science, and drug discovery.

Book a demo