Patentability Search Process: Five Practical Steps
A repeatable five-step workflow from receiving a technical disclosure to producing a reviewable patentability search report.
The patentability search process can feel complex: with extensive patent and non-patent literature, where should a searcher begin, what should each step produce, and how should uncertainty be documented?
This article breaks the patentability search process into five actionable steps, from receiving a technical disclosure to preparing a source-linked report. The method is useful for newcomers and experienced professionals because it separates understanding, retrieval, legal comparison, and reporting instead of treating search as one undifferentiated task.
No search can guarantee that every relevant disclosure has been found. A strong process instead uses multiple retrieval routes, records its scope and limitations, and creates clear feedback loops when new terminology or evidence changes the analysis.
- Five steps: understand the invention, build a strategy, search multiple sources, compare features, report.
- Novelty is assessed one reference at a time. Inventive step is a separate analysis under a named jurisdictional test.
- All five steps can run automatically: Eureka IP Novelty Search extracts the features, builds the search strategies, and returns a feature-by-feature comparison report.
Overview: the five-step patentability search process
- Understand the invention: identify the technical problem, solution, effects, and potentially distinguishing features.
- Develop a search strategy: construct keyword, classification, citation, and document-based search routes.
- Execute the search: search appropriate patent and non-patent sources and refine iteratively.
- Comparative analysis: map disclosures to claim-relevant technical features and separate novelty from inventive-step analysis.
- Prepare the report: document scope, evidence, conclusions, uncertainty, and recommended next review.
These steps are iterative rather than strictly linear. Patsnap Eureka IP Search includes a Novelty Search workflow that extracts technical features, builds multiple search strategies, and compares close prior art feature by feature.1 Its output can organize evidence for professional review, but it cannot guarantee completeness or determine legal patentability.
Run a novelty search on your own invention
Paste the technical disclosure and the agent works through all five steps below, from feature extraction to a source-linked comparison report. You can review or edit every step before it runs.
Step one: understand the invention, the foundation of patentability searching
Why this step is easy to overlook
A search can be technically well executed and still answer the wrong question. If the searcher misunderstands the invention, later queries may overemphasize a product label, miss the real technical contribution, or compare the wrong features.
Read the technical disclosure
Review the disclosure from four dimensions:
- Technical problem: what practical limitation is being addressed, and how do existing approaches fall short?
- Technical solution: what components, steps, relationships, materials, parameters, or controls produce the result?
- Technical effect: what measurable or technically credible effect follows from the distinguishing features?
- Embodiments: what examples, drawings, test data, and alternatives explain how the solution works?
Communicate with the inventor
Where possible, ask focused questions:
- What is the most fundamental technical difference from known solutions?
- Which feature would you keep if the invention had to be described in one sentence?
- Which assumptions seem obvious to the team but may not be obvious to an outside searcher?
- Which alternatives and failed approaches help define the technical boundary?
Distill a technical feature checklist
| Feature | Example description | Role | Searchable concepts |
|---|---|---|---|
| F1 | Material A used as a thermally conductive medium | Potentially essential | Material family, composition, thermal function |
| F2 | Medium falls within a defined thickness range | Preferred or claim-dependent | Units, endpoints, ranges, layer terms |
| F3 | Structure B increases heat transfer | Potentially essential | Geometry, relationship, functional language |
The feature list should distinguish observable technical content from subjective outcomes. “Better user experience” is not a useful search unit unless the underlying technical mechanism is stated.
Mistakes to avoid: relying on the title alone, treating an inventor’s conclusion as a technical feature, ignoring alternative embodiments, and failing to identify which effects are actually supported by the disclosure.
Step two in the patentability search process: develop the search strategy
Keyword strategy
The same concept may appear under different terminology across industries, languages, and drafting styles. Build a concept table for each core feature:
- Technical terms: terminology used by engineers, standards, and patent drafters.
- Functional terms: the action or result produced by the feature.
- Broader and narrower concepts: material classes, specific embodiments, components, and substitutes.
- Translations: English, Chinese, and other target-market languages, checked for technical meaning rather than translated mechanically.
For example, a heat-dissipation concept might require “heat dissipation,” “thermal management,” “cooling,” “heat transfer,” and application-specific terminology. The exact set should follow the technology, not a generic synonym list.
For related context, Patsnap’s discussion of AI-assisted novelty analysis explains how semantic retrieval can complement keyword-led searching.
Classification strategy
Classification searches reduce dependence on wording. The International Patent Classification IPC is a hierarchical, language-independent system for classifying patents and utility models by technology.2 CPC may provide additional granularity for some collections, while Japan’s FI and F-term systems provide further search routes for Japanese documents.3
- Run a broad keyword search and identify several clearly relevant documents.
- Review their classifications and the definitions above and below the assigned group.
- Confirm that the classification captures the relevant technical concept rather than merely the product context.
- Combine classifications with keywords, dates, applicants, citations, or other fields as appropriate.
Construct search queries
Use the syntax supported by the selected database. Common Boolean concepts include:
- AND to require multiple concepts;
- OR to combine synonyms and variants;
- NOT cautiously, because exclusions can remove relevant documents; and
- proximity, truncation, wildcards, and field restrictions only according to the database’s documented syntax.
A conceptual query might combine a thermal-management set, a carbon-material set, and a relevant classification. Validate every character and wildcard in the target system; query syntax is not portable across all databases.
Document the strategy
Record the search date, searcher, target jurisdictions, relevant date, databases, language coverage, keywords and rationale, classification codes and definitions, full query strings, filters, result counts, and screening notes. These records improve repeatability and explain what the search did and did not cover; they should not be presented as proof that the search is complete.
Let the agent build this search strategy for you
It returns keyword sets, IPC and CPC codes, and semantic, Boolean, citation-tracking and non-patent routes. Edit any of them, then run the search.
Step three in the patentability search process: execute the search
Select sources by purpose
| Source | Verified role | Practical use | Access |
|---|---|---|---|
| USPTO Patent Public Search4 | U.S. patents and published applications | U.S.-focused keyword and field searching | Public |
| Espacenet5 | Worldwide patent information with search and family-related tools | Broad international patent retrieval and document review | Public |
| WIPO PATENTSCOPE6 | Published PCT applications and participating national or regional collections | PCT and multilingual searches, including IPC and other fields | Public |
| J-PlatPat3 | Official Japanese digital library and FI/F-term routes | Japanese patent and utility-model searching | Public |
| Google Patents7 | Patent publications plus indexed technical material, with stated coverage limitations | Fast supplementary discovery and family or citation exploration | Public |
| Derwent Patent Search (formerly Derwent Innovation)8 | Commercial patent-search software with curated summaries and multiple search modes | Professional searching where licensed data and indexing fit the matter | Commercial |
This is an illustrative, non-exhaustive source list, not a ranking. Coverage, syntax, update timing, family grouping, translations, and access conditions differ. Check each source’s current documentation and use additional national, scientific, standards, product, or archival sources when the technology requires them.
Patsnap’s patent analytics overview discusses semantic search, citation analysis, and other tools that may complement multi-source review.
Execute iteratively
- Begin with broad concepts and a source suited to the target jurisdictions.
- Review highly relevant documents for terminology, classifications, citations, inventors, applicants, and family members.
- Refine or branch the query based on what the documents reveal.
- Search non-patent literature where public technical disclosure may be material.
- Record each meaningful query and screening decision.
Do not use universal hit-count thresholds such as “more than 1,000 is too many” or “fewer than 50 is too few.” A useful result set depends on field size, database behavior, family grouping, query purpose, and screening capacity.
Step four in the patentability search process: compare prior art
This step turns retrieved documents into a structured legal and technical review. Select the most relevant references based on what they disclose, not a fixed quota.
Build the comparison matrix
| Claim-relevant feature | D1 | D2 | Non-patent document | Notes |
|---|---|---|---|---|
| Material A as thermal medium | Expressly disclosed | Not identified | Material family discussed | Check whether variant A′ falls within the claim wording |
| Specified thickness range | Different range | Overlapping value | Not identified | Analyze the exact disclosure and applicable range law |
| Structure B for heat transfer | Not identified | Expressly disclosed | Not identified | Do not combine references for novelty |
Separate novelty and inventive step
For novelty, evaluate one prior-art item at a time. The EPO Guidelines state that separate items of prior art may not be combined for novelty.9 The precise anticipation standard depends on the governing jurisdiction and claim construction.
If the claim appears novel, conduct a separate inventive-step or non-obviousness analysis. Under the EPO problem-solution approach, this involves the closest prior art, the objective technical problem, and whether the claimed invention would have been obvious.10 Other jurisdictions apply their own frameworks, so label the legal standard used.
A comparison table should distinguish “expressly disclosed,” “implicitly disclosed,” “not identified,” and “requires legal interpretation.” A blank cell is not proof that a feature is absent.
Generate this comparison table automatically
The agent maps every feature against each close reference, tags the references PS-X, PS-Y or PS-A, and links each mapping back to the passage it came from.
Step five: prepare the report, findings, limitations, and next review
Elements of a patentability search report
- Search summary: dates, searcher, purpose, jurisdictions, relevant date, sources, queries, languages, classifications, and limitations.
- Invention summary: the technical problem, solution, effects, and feature list used for searching.
- Key references: publication numbers, dates, titles, applicants or assignees where relevant, and stable source links.
- Feature comparison: the matrix and the passages supporting each disclosure mapping.
- Novelty analysis: one-document comparison under the selected jurisdictional rule.
- Inventive-step analysis: a separate, reasoned analysis under the applicable framework.
- Overall assessment: identified risks, unresolved questions, search limitations, and issues for counsel, not a guaranteed grant prediction.
- Recommendations: additional searching, technical clarification, supporting data, claim-development options, or professional review.
Once the search record is reviewed, Patsnap Eureka IP Drafting offers separate invention-disclosure, patent-drafting, and office-action-response workflows and supports CNIPA, USPTO, and EPO standards.11 The search report should inform drafting, not dictate filing, abandonment, or trade-secret decisions without legal and commercial review.
Legal-information notice: patentability search results and AI-generated outputs may be incomplete or inaccurate. They are not legal advice and require independent professional review before filing or acting.12
Feedback loops: patentability searching is not linear
- New terminology discovered during searching → return to step two and expand the concept table.
- A new technical field appears during comparison → return to step three with a targeted search branch.
- A disclosure mapping remains uncertain → return to step one for technical clarification or step four for closer reading.
- A report conclusion depends on an unsupported effect → request evidence or narrow the conclusion.
The goal is not subjective “complete confidence.” The goal is a defensible stopping point based on documented scope, diminishing new information from additional search routes, the decision at stake, and the review resources available.
Illustrative case: applying the five-step methodology
Scenario: a startup develops a foldable bicycle helmet.
Step one, understand the invention: the disclosed concept uses multiple curved shell segments and hinge relationships to move between a wearable configuration and a compact configuration. The searcher asks the inventor to clarify load transfer, locking, safety behavior, and which geometry is essential.
Step two, develop a strategy: the concept table includes “foldable helmet,” “collapsible protective headgear,” “segmented shell,” “hinged shell,” locking structures, portability functions, and verified relevant classifications. Classification codes are confirmed in the current classification scheme rather than copied from an example.
Step three, execute the search: the searcher uses international and target-jurisdiction sources, reviews patent families and citations, and searches non-patent material for public product disclosures.
Step four, compare: one reference discloses a hinge-connected folding structure, while another discloses curved shell segments. Neither finding alone answers the legal question. Novelty is assessed against each reference separately; inventive step is analyzed under the target jurisdiction’s framework.
Step five, report: the report identifies the closest documents, maps each feature to evidence, explains unresolved obviousness risk, and recommends that counsel evaluate claim scope around the specific segment geometry, hinge relationship, locking behavior, and supported technical effects.
This scenario is illustrative. The document counts, classification choices, patent outcome, and legal conclusion in a real matter would depend on the evidence, claim language, relevant date, jurisdiction, and search scope.
Run an FTO search on the same product
Once the design stops moving, the question changes from patentability to clearance. Reuse the technical features to screen live claims for infringement risk and get claim charts in minutes.
Key takeaway
The patentability search process follows five core steps: understand the invention, develop a search strategy, execute a multi-source search, compare claim-relevant features, and prepare a transparent report. Its quality comes from documented reasoning and iterative refinement, not from an unsupported promise that nothing was missed.
Every conclusion should trace back to a feature, a cited passage, a relevant date, and a stated jurisdictional test.
Sources and verification
- Patsnap Eureka, AI Patent Search, FTO & Design Clearance. Accessed July 28, 2026.
- WIPO, International Patent Classification. Accessed July 28, 2026.
- Japan Patent Office, Patent and Utility Model Search (J-PlatPat). Accessed July 28, 2026.
- USPTO, Patent Public Search. Accessed July 28, 2026.
- European Patent Office, Espacenet. Accessed July 28, 2026.
- WIPO, PATENTSCOPE. Accessed July 28, 2026.
- Google Patents, Coverage and limitations. Accessed July 28, 2026.
- Clarivate, Derwent Patent Search. Accessed July 28, 2026.
- EPO Guidelines G VI, 1: Novelty and state of the art. Accessed July 28, 2026.
- EPO Guidelines G VII, 5: Problem-solution approach. Accessed July 28, 2026.
- Patsnap Eureka, AI Patent Drafting Assistant. Accessed July 28, 2026.
- Patsnap Eureka, Terms of Service: AI output limitations. Accessed July 28, 2026.
Official patent-office, WIPO, product, and database documentation was reviewed in July 2026. Database coverage, query syntax, classifications, and product capabilities can change.
Build a reviewable search record
Turn technical disclosure into structured features, search strategies, source-linked prior art, and a comparison report ready for professional review.
Start a novelty search
