Patent attorney working methods: from a research question to argument and evidence

Professional methods, legal analysis and scientific research

Patent attorney work brings together technical language, legal rules and procedural organisation. No single method suits every matter: a technical solution, a brand name and a product appearance call for different tests. This article explains analytical methods useful in this field. The sequence of steps and suggested working documents are Rekel Patent’s own editorial synthesis, rather than an official professional standard or a description of services offered by its publisher.

Three tasks need to be distinguished. Legal analysis identifies the meaning of rules and applies them to facts. Technical analysis compares solutions, mechanisms and effects. Scientific research additionally requires a defined question, transparent material-selection criteria and a method by which results can be assessed. Aristodemou and Tietze review AI and machine learning applications in IP data analysis; that work does not replace the legal tests in an individual matter. ^[11]

1. Defining the question and the analytical boundaries

The starting point is a clear question: can a solution be filed for protection, is a name registrable, might a product infringe another party’s right, or are there grounds for an opposition? In the proposed method, the question is recorded alongside territory, assessment date, product version and intended business decision. “Check the patent” is too broad: it may mean patentability, legal status or freedom to operate.

A map of the relevant subject matter follows: technical solution, name, logo, product appearance, documentation and confidential information. One product may require several separate analyses. For a Kraków team planning sales in Poland and Germany, this includes separating the destination market from where the project was created. This is an illustration of how to organise analysis, not an assessment of a particular product.

2. Chronology, creators and the chain of rights

Before searching earlier rights, it is useful to organise the project’s own records: versions, tests and disclosures, contributors, contracts, employment scope, university collaboration and funding. WIPO describes different models for rights in research results, depending on the legal system and relationship with the creator. Funding a project alone does not resolve every question about entitlement to file for protection. ^[1]

The editorial proposal is a table with five columns: event, date, document, person and unresolved issue. It separates a documented fact from a statement or assumption. For a trade-show disclosure, photographs of the stand are only part of the picture: what was actually shown, to whom and under which conditions must also be established. A convenient date must not be presented as certain simply because it supports the preferred argument.

3. Search strategy: words, IPC/CPC and patent families

Espacenet supports patent-publication searches and technical classification, with tools for accessing file wrappers made available by patent offices. A strategy should go beyond a product’s marketing name. A practical plan may combine functional terminology, synonyms, inventor and applicant names, and IPC/CPC symbols, then examine related and cited documents. ^[2]

The proposed search log records the database, date, complete query, filters, time range, number of results and selection criteria. Patent families are organised so that multiple publications in one family are not casually counted as independent inventions. Keyword and classification searches play complementary roles. Differences between their results prompt refinement of the strategy, rather than selection of the more convenient dataset. This is a proposal for transparent reporting of the work.

4. Novelty and inventive step: two separate tests

Under the EPO Guidelines, a novelty assessment does not combine separate prior-art disclosures into an artificial earlier “invention”. What each item discloses must be established. A working matrix can match each claim feature to a document passage and its availability date. An empty field needs explanation; a similar title or illustration does not establish disclosure of every feature. ^[3]

For inventive step, the EPO uses the problem–solution approach: identify the closest prior art, establish the objective technical problem, and assess whether the claimed solution would have been obvious to the skilled person. The analysis therefore explains the differences and their technical effects, rather than merely saying “our product is better”. This structure makes the reasoning explicit without attributing knowledge from the application under examination to an earlier inventor. ^[4]

5. FTO: patentability and freedom to operate answer different questions

A patentability assessment concerns whether protection can be obtained for one’s own solution. Freedom to operate (FTO) addresses potential obstacles arising from other parties’ rights when carrying out the planned activity. WIPO’s guide explains searching and analysing published documents for this purpose. The product must be linked to claim language and the status of the right in the relevant territory. Holding one’s own patent does not automatically establish freedom to use the solution. ^[5]

The proposed report separates finding a document, interpreting scope and reaching a risk conclusion. For example, a new device component may merit protection of its own while the device as a whole uses an earlier protected solution. The conclusion identifies the product version, countries and date; a design change or entry into another market triggers reconsideration. ^[5]

6. Trade marks: distinctiveness and the global assessment of conflict

A name analysis separates absolute grounds for refusal from potential conflicts with earlier rights. EUIPO emphasises that refusal grounds are independent: descriptiveness and lack of distinctive character each need their own reasoning even where they overlap. The question is therefore both “does someone already have this name?” and “how will the public understand it for these goods or services?”. ^[6]

Likelihood of confusion involves the goods and services, relevant public, level of attention, signs and distinctiveness of the earlier mark. The global assessment considers how these factors interact; spelling similarity is only part of the analysis. A visual, phonetic and conceptual matrix helps organise arguments, but adding numerical scores cannot replace legal reasoning. ^[7]

7. Designs: comparing the overall impression

For a design, novelty and individual character are also separate questions. EUIPO explains that identity includes differences confined to immaterial details. Proper comparison material is therefore needed, rather than two arbitrarily chosen photographs taken from different angles. ^[8]

Individual character involves the informed user, the designer’s freedom and the overall impression. Appearance features dictated solely by technical function must also be distinguished from choices relevant to the design assessment. The editorial documentation proposal uses comparable views, marked differences and a separate field for functional constraints. It is not a percentage image-matching test. ^[9]

8. Claims, evidence and the procedural calendar

Article 84 EPC requires claims to define the matter for which protection is sought and to be clear, concise and supported by the description. Drafting therefore links terminology to the actual disclosure. A working matrix can connect a feature, its basis in the description, a drawing, an embodiment and its technical rationale. This organisation does not determine whether a later application amendment would be admissible. ^[10]

The proposed case organisation also uses an “assertion — fact — evidence — counterargument” matrix and an action calendar. Each deadline is recorded with its basis, starting event, calculation, responsible person and completion confirmation. Official deadlines, internal working dates and business milestones must be distinguished. Evidence review addresses provenance, completeness, date and reproducibility; a file list with no link to specific assertions is insufficient.

9. Scientific method, AI and the limits of inference

For a research article or technology landscape, the protocol should disclose the question, unit of analysis, inclusion criteria, coding, deduplication and dataset limitations. IP analytics literature discusses tools for discovering patterns in data. A similarity ranking or patent count does not independently answer questions about validity, infringement or technological quality. ^[11]

Rekel Patent’s editorial rule is that AI may assist with vocabulary, grouping results or drafting a preliminary summary, but every publication number, quotation, date and central conclusion returns to its source document. The report ends with gaps and conditions for updating it. Finding no document is the result of a specified search, not proof that an earlier right does not exist. None of these methods guarantees a complete search or a procedural outcome.

Sources and further reading

1. 1
   WIPO, FAQs: Technology Transfer — IP ownership ↗

   WIPO, FAQ: IP ownership

   Different models of entitlement to work results and the importance of the applicable law.

2. 2
   EPO, Espacenet — patent search ↗

   EPO, Espacenet documentation

   Patent-publication searches, classifications and access to file wrappers.

3. 3
   EPO Guidelines 2026, G-VI, 1 — Novelty and state of the art ↗

   G-VI, 1; EPC Art. 54

   Novelty and the prohibition on combining separate disclosures for this assessment.

4. 4
   EPO Guidelines 2026, G-VII, 5 — Problem-solution approach ↗

   G-VII, 5; EPC Art. 56

   The stages of the problem–solution approach for inventive step.

5. 5
   WIPO (2020), Identifying Inventions in the Public Domain: A Guide for Inventors and Entrepreneurs ↗

   WIPO Publication No. 1062

   Searching and analysing patent documents for freedom-to-operate determination.

6. 6
   EUIPO Guidelines, Part B, Section 4, Chapter 1, 1 — Reasoned Objection ↗

   EUIPO Guidelines, edition 2026; EUTMR Art. 7(1)

   Separate assessment of absolute refusal grounds, including lack of distinctiveness and descriptiveness.

7. 7
   EUIPO Guidelines — Global assessment of likelihood of confusion ↗

   EUIPO Guidelines, edition 2026, Part C, Section 2, Chapter 7, 1

   Interdependence of the similarity of signs, goods or services and other factors.

8. 8
   EUIPO Design Guidelines, 5.7.2.1 — Novelty ↗

   EUIPO Design Guidelines, edition 2026, 5.7.2.1

   Design novelty and immaterial differences.

9. 9
   EUIPO Design Guidelines, 5.7.2.2 — Individual character ↗

   EUIPO Design Guidelines, edition 2026, 5.7.2.2

   The informed user, designer freedom, overall impression and functional features.

10. 10
    European Patent Convention, Article 84 — Claims ↗

    EPC, Article 84

    Claims define the matter protected and must be clear, concise and supported by the description.

11. 11
    Aristodemou, L., Tietze, F. (2018). The state-of-the-art on Intellectual Property Analytics (IPA): A literature review on artificial intelligence, machine learning and deep learning methods for analysing intellectual property (IP) data ↗

    DOI: 10.1016/j.wpi.2018.07.002; World Patent Information 55, 37–51

    Literature review of AI, machine learning and deep learning analysis of IP data, distinct from individual legal assessment.