Legal basis for inventive step assessment

Under EPC Article 56, an invention involves an inventive step if, having regard to the state of the art, it is not obvious to a person skilled in the art.¹ Article 54(3) prior art is excluded from that inventive-step assessment.

The skilled person is a legal construct: an average practitioner in the relevant field who knows the common general knowledge at the relevant date, has access to the state of the art, and has the ordinary means and capacity for routine work and experimentation in that field. Depending on the problem, the construct may represent a team rather than one individual.²

The EPO applies the problem-solution approach to make the assessment more objective and predictable. Its three stages are selecting the closest prior art, establishing the objective technical problem, and asking whether the claimed invention would have been obvious from that starting point and problem.³

A repeatable analysis begins with reliable retrieval and feature mapping. Eureka IP Search’s Novelty Search extracts core technical features, builds multiple search strategies, and compares close prior art feature by feature.⁴ That evidence can inform the starting-point analysis, but the tool does not make the legal inventive-step determination.

Problem-solution approach step 1: determine the closest prior art

The starting point anchors the later analysis. It should be a realistic or feasible springboard for assessing whether the skilled person would arrive at the claim, not a document selected only because it makes the invention appear more or less inventive.

What makes a starting point relevant?

The 2026 EPO Guidelines describe the closest prior art as a single reference containing the feature combination that provides the most promising starting point. The first consideration is a similar purpose or effect, or at least the same or a closely related technical field. In practice, it often concerns a similar use and requires relatively few structural and functional modifications.⁵

Common starting-point mistakes

• Selecting an artificially weak document: choosing a reference because it discloses few features makes the analysis less persuasive.
• Automatically excluding another field: the skilled person may look to neighbouring, general, or even remote fields when the objective problem provides a prompt.
• Treating “closest” as a feature-count contest: purpose, effect, technical context, and the modification path matter alongside overlap.
• Running unlimited attacks: several starting points may be valid, but separate analyses are required only when they are convincingly shown to be viable springboards. One relevant starting point can be enough to support a lack-of-inventive-step conclusion.⁵

Problem-solution approach step 2: identify differences and the objective technical problem

Identify the distinguishing technical features

Compare the claim with the selected starting point feature by feature. Record the structural and functional differences, including relationships, operating conditions, materials, sequence, and claimed technical functions. A technical effect is not itself automatically a distinguishing feature; first connect the effect to one or more claimed differences.

Determine the technical effects of the differences

Ask what technical effect each distinguishing feature produces in the context of the claim. The effect should be supported by the application and credible across substantially all embodiments relied on for that problem. An effect need not be word-for-word in the application, but it must be encompassed by the original technical teaching and embodied by the originally disclosed invention.

Formulate the objective technical problem

The objective problem is the task of modifying or adapting the starting point to achieve the effects linked to the differences. It may differ from the problem stated by the inventor because the prior art found during the search can change the factual baseline. It must not contain a pointer to the claimed solution.⁶

If the asserted improvement is not credible across the relied-on scope, use a narrower supported effect or formulate the problem as providing an alternative. When separate groups of features solve independent partial problems, assess those groups separately rather than inventing a single combined effect.

Before formal assessment, Eureka IP Drafting’s Invention Disclosure workflow can turn rough engineering notes into structured disclosure and highlight missing details or overlooked inventive points.⁷ This adjacent intake workflow can help clarify the claimed contribution; it does not establish the objective technical problem or decide inventive step.

Problem-solution approach step 3: assess obviousness

The EPO’s could–would question is not whether the skilled person could have modified the starting point. It is whether the prior art as a whole would have prompted that person, when facing the objective problem, to make the modification and arrive at the claim in expectation of an improvement or advantage.⁸

Scenario 1: another reference teaches the feature for a relevant purpose

Suppose D1 uses a metal heat sink and the claim replaces it with a graphene thermal film. D2 may disclose graphene for heat dissipation in electronic devices. The analysis still must explain why the skilled person would consult D2, apply its teaching to D1, preserve necessary functions, and arrive at the claimed arrangement. A matching feature in D2 is evidence, not a complete obviousness conclusion.

Scenario 2: the feature is asserted to be common general knowledge

Common general knowledge may supply a prompt or routine implementation detail, but the assertion must be supportable. Basic textbooks and monographs can represent such knowledge; a single patent or journal article normally does not, subject to exceptions for new or specialized fields.⁹ If the status is disputed, identify documentary evidence and the relevant date instead of simply labelling the feature conventional.

Scenario 3: routine work or experimentation

The skilled person has ordinary capacity for routine work and experimentation. Still, “limited experimentation” is not an independent rule that makes a claim obvious. The analysis should identify the prior-art prompt, the predictable path, the available alternatives, and the expectation of success. If the problem leads step by step to the solution and each step is obvious in light of what has already been achieved, the overall route may be obvious.⁸

Circumstances that may point away from the combination

• a reference discourages the proposed modification or creates a technical prejudice;
• the relevant disclosures contain essential features that are technically incompatible;
• the path requires additional choices or redesign for which the prior art supplies no prompt;
• the claimed combination produces a precisely stated, supported, and genuinely unexpected technical effect.

Combining more than one reference is permitted in EPO inventive-step analysis. The number alone does not decide the issue. The reviewer should explain why the skilled person would associate the teachings and address field, compatibility, partial problems, and any clear cross-reference.¹⁰

Practical case study of the problem-solution approach

Illustrative claim: an electric-vehicle battery thermal-management method activates liquid cooling above a first temperature threshold and also activates auxiliary air cooling above a higher second threshold.

Step 1: select a feasible starting point

D1 discloses battery thermal management for an electric vehicle using liquid cooling that activates when battery temperature becomes too high. D1 has a similar purpose and use, and it already discloses part of the claimed control strategy.

Step 2: define the difference, effect, and objective problem

• Distinguishing feature: a second, higher threshold triggers auxiliary air cooling in addition to liquid cooling.
• Proposed technical effect: extra cooling capacity becomes available at a higher thermal load.
• Objective technical problem: how to provide additional cooling capacity when the thermal load exceeds the capacity addressed by the first cooling mode.

The asserted effect must be supported. Do not add “preventing failure” or a quantified improvement unless the application and evidence support that result across the claim.

Step 3: apply could–would

Assume D2 concerns server thermal management and teaches combined liquid and air cooling under high thermal load. The relevant questions are whether the objective problem would prompt the skilled person to consult that field, whether D2’s teaching applies to the battery context, whether the systems are compatible, and whether the claimed two-threshold control follows without hindsight.

If D2 expressly links the combined modes to the same type of peak-load problem, the hardware can be transferred without incompatible changes, and the skilled person would expect the arrangement to work, the claim may be obvious. If the transfer requires an unprompted control architecture, addresses materially different safety constraints, or produces a supported unexpected interaction, inventive step may remain.

Possible supporting points include a demonstrated synergistic cooling effect tied to the claimed features or an implementation that achieves the effect without hardware that D2 treats as necessary. These are evidentiary questions, not automatic escape routes.

Secondary indicators in inventive-step assessment

Secondary indicators can support the overall analysis, but they do not mechanically reverse a result reached under the three-step framework.

Unexpected technical effect

An unexpected effect can indicate inventive step when it results from the claimed subject matter and is precisely stated. A vague assertion of unexpectedly good performance is insufficient. Moreover, if the skilled person would already have reached the claim through a one-way-street situation, the effect may be only a bonus effect.¹¹

Overcoming technical prejudice

Evidence that the skilled community generally expected a route not to work can weigh against obviousness. The prejudice should be shown with dated technical evidence, not inferred from the absence of earlier adoption.

Long-felt need and commercial success

A long-standing technical problem that workers repeatedly attempted to solve may indicate inventive step. Commercial success alone does not. It becomes relevant when paired with evidence of a long-felt need and when the success is attributable to the claimed technical features rather than advertising, sales methods, or other factors.¹²

Common mistakes in inventive-step assessment

Mistake 1: hindsight bias

Once the invention is known, a sequence of easy-looking steps can often be reconstructed from the prior art. EPO guidance warns against this ex post facto analysis. Rebuild the state of the art as it confronted the skilled person before the effective date, state the objective problem without solution language, and identify the contemporaneous prompt for each modification.¹³

Mistake 2: fragmenting the prior art without a combination rationale

A combination of D1, D2, D3, and common general knowledge is not automatically prohibited. Nor is it automatically persuasive. For each added teaching, explain the problem it addresses, why the skilled person would consult it, how the teachings fit, and whether the route remains predictable. Multiple independent partial problems may justify different supplementary references.

Mistake 3: treating an unexpected effect as independently decisive

An effect must be linked to the claimed characterising features, precisely described, supported, and relevant across the relied-on claim scope. It is an indicator within the inventive-step analysis, not a substitute for defining the starting point, differences, objective problem, and could–would reasoning.

Key takeaways

The problem-solution approach has three connected stages: choose a feasible prior-art starting point, identify the claimed differences and formulate a non-leading objective technical problem, then ask whether the prior art would—not merely could—have prompted the skilled person to reach the claim.

A reviewable patentability analysis records the starting-point rationale, feature mapping, supported technical effects, problem formulation, combination logic, expectation of success, and evidence for any secondary indicator. The result remains a preliminary assessment, not a legal opinion on validity, infringement, or freedom to operate.