Patent Search and Patentability Analysis: Evidence-Based Research Framework — research analysis
Begin with a disclosure that can be searched
A search request should identify the technical problem, essential features, differences from known solutions and variants genuinely contemplated. A product description such as AI platform or medical device is too broad to support a claim-focused search. Break the invention into components, functions, interactions and measurable outcomes. Mark mandatory features separately from optional preferences. Record inventors' terminology but expand it using alternative technical terms. This initial claim-feature map helps avoid spending the entire search on an attractive keyword unrelated to the eventual claims.
Use a reproducible query and classification strategy
Search patent databases using synonyms, neighboring technologies, known applicants, citations and classification systems. WIPO's IPC provides a hierarchical technology taxonomy; the USPTO's public patent search provides official U.S. document access. Begin with broad queries to learn the field, then move to structured combinations designed to test individual claim elements. Do not assume that one patent class contains every relevant publication. Record the exact query string, date, filter, database and retrieved publication number so another analyst can repeat the search. WIPOIPC USSEARCH
Include non-patent literature and language variants
A patentability research process is not limited to granted patents. Published applications, academic articles, standards, theses, preprints, technical manuals and publicly accessible product information can matter. Earlier non-English documents may also disclose the claimed features; translations should be checked against the original where precision matters. Do not conflate a search platform's publication date with a patent's effective filing or priority date. Maintain a source register specifying what each document actually teaches, whether its date is verified and whether the relevant information was publicly available in the applicable period.
Distinguish anticipation from combinations of disclosures
A novelty assessment commonly asks whether a single relevant prior-art disclosure teaches the required claim elements, while inventive-step or obviousness analyses may consider additional teachings and their relation to one another under the applicable law. The USPTO's MPEP discusses enabling prior-art disclosure and how technical operability issues affect rejections. A report that lists ten similar patents without showing element-by-element relationships is weaker than a chart of two or three truly close references. Analyze what the reference says, not merely what its title suggests. USPRIOR
Make search limitations explicit
No finite search establishes that every possible disclosure worldwide has been found. Databases differ in coverage, indexing, publication lag, language availability and full-text quality. Some applications remain unpublished, and non-patent knowledge may be difficult to access. State the search date, period, databases, languages, classifications and excluded material. Flag unresolved claim terms or incomplete inventor evidence. A research rating should explain whether the conclusion is based on direct disclosure, inference or an unverified possibility rather than offer a single unexplained green/red patentability label.
Do not confuse patentability with freedom to operate
A patentability search examines whether a proposed claim may be new and inventive; FTO research concerns active third-party claims that could cover a contemplated activity in a territory. A patent can be granted while another unexpired patent constrains manufacture or sale of the resulting product. The analytical search and legal clearance work therefore require different search universes and different outputs. A patentability report should not be presented as an infringement opinion. If the client's real decision concerns launching a product, explicitly identify the need for a separate territorial FTO review.
Deliver a decision-ready search report
The report should contain the invention definition, query log, patent families, relevant non-patent documents, priority/date notes, a claim-feature matrix, novelty and inventive-step questions, confidence limitations and research next steps. For each critical reference, link the source and cite the particular paragraph, drawing or claim. Add alternative claim directions only as hypotheses supported by inventor evidence. A final section should identify what additional testing or expert review could change the conclusions. This allows inventors and patent practitioners to make better-informed choices while preserving the limits of the research assignment.
How to document findings so they remain auditable
A usable patentability search log should contain the exact technical disclosure searched, keyword variants, IPC/CPC classes, date filters and search history. Attach each important patent publication as a specific family record, with independent confirmation of priority and publication dates. When citing a relevant paragraph or figure, copy its location into a claim-feature chart so the reader can see the connection. For non-patent sources, preserve stable bibliographic details and evidence of public availability. State coverage limitations, including languages and search dates, and do not describe a list of superficially similar abstracts as a completed novelty opinion. A particularly useful synthesis is an evidence-to-claim table with a status per limitation: disclosed in one reference, potentially suggested by a combination, absent from reviewed records, or not yet searchable because the invention is underspecified. Provide the exact source location behind each status. This makes uncertainty visible and avoids a misleading overall score.
A useful research conclusion needs to distinguish a source fact from the analyst's interpretation. Patent records, office guidance, legislation and technical publications can establish what was disclosed or required at a given time; they cannot, without further analysis, establish a specific invention's patentability or grant probability. Preserve source dates, the material claim or technical feature, jurisdiction, and contrary evidence in the same record. If any field is unknown, show it as unresolved. The scope of research can be described accurately without claiming a legal opinion was issued.
Decision framework: evidence, uncertainty and next action
The next decision should be driven by the closest relevant references and the available disclosure, not the raw number of retrieved results. If an element is already taught, investigate the remaining technical combination; if the disclosure lacks implementation detail, obtain inventor evidence. A market-launch concern calls for a separate FTO study. Avoid collapsing these outputs into a single 'patent clear' conclusion.
Research control matrix
| Evidence question | Required record | Interpretation limit |
|---|---|---|
| What feature is distinctive? | Inventor-supplied technical description and dated evidence | Not a patentability determination |
| Which earlier disclosures matter? | Publication numbers, dates and precise passages | Search coverage is limited |
| Which jurisdiction is being considered? | Current official rule, filing system, claim category | Other territories may differ |
| What could change the analysis? | Missing experiment, unresolved claim term, contrary reference | Flag before filing or launch |
Research next steps
The research workstream should identify the most material unresolved questions, update the source register where rules or technical evidence change, and prepare the relevant documentation for an appropriate specialist. A follow-up may involve a deeper prior-art search, evidence comparison, supporting experimental records or a separate legal analysis. The objective is a traceable decision rather than a confident answer unsupported by the particular facts.
Official and primary references
Source scope and publication dates should be checked before an invention-specific decision.
- USSEARCH — https://www.uspto.gov/patents/search/patent-public-search
- USPRIOR — https://www.uspto.gov/web/offices/pac/mpep/s2121.html
- WIPOIPC — https://www.wipo.int/classifications/ipc/en/
- EPOSEARCH — https://www.epo.org/en/searching-for-patents/technical/espacenet
- US112 — https://www.govinfo.gov/content/pkg/USCODE-2024-title35/html/USCODE-2024-title35-partII-chap11-sec112.htm

How to patent fintech inventions
Gen AI Innovations
What is Extended Reality
How to patent artificial intelligence based innovations
Patent claim drafting