Analysis Methodology
Evalify runs a structured, multi-stage methodology that mirrors a disciplined FTO workflow. Understanding the stages helps you interpret the output and judge where your own review adds the most value.
1 · Subject-matter decomposition
Evalify parses your input and decomposes the subject matter into distinct technical aspects, each expressed in claim-style language. This framing — treating the product as a set of concrete, chartable features rather than a narrative — is what lets the downstream analysis be specific.
You can review and refine these aspects before the full analysis runs. Curating them (adding, removing, or sharpening aspects, or seeding key concepts) is the single highest-leverage way to focus the search and charting on what matters for your matter.
2 · Search
For each aspect, Evalify runs a semantic and classification-based search across a comprehensive, licensed global patent corpus, scoped to the jurisdictions you selected. It combines conceptual (semantic) matching with classification signals to build a candidate set, then screens for relevance so the analysis focuses on references that plausibly read on the subject matter rather than an unfiltered result list.
Because freedom to operate is a per-market question, Evalify also runs a dedicated search of each selected market to make sure none is under-covered — a real concern for non-English offices (Japan, Korea, China), whose documents are easy to miss in a single combined ranking. This is what lets a market read as genuinely clear "as searched" rather than clear merely because its patents ranked low. See Clearance by Jurisdiction.
3 · Claim retrieval and construction
For the relevant candidates, Evalify retrieves and normalizes the independent claims — the operative claims for an infringement read — as the basis for charting. Claim language in other source languages is handled so the analysis works over a consistent, English-language claim set.
4 · Claim charting — the core
This is the heart of the analysis. For each aspect, Evalify constructs a limitation-by-limitation claim chart against the relevant claims: it maps each claim limitation to the corresponding feature of the subject matter and assesses whether it is present. This all-elements framing — rather than a holistic "similarity" judgment — is what makes the output defensible and lets you see exactly which limitation drives (or defeats) a potential read, and where the subject matter is clear.
5 · Legal status, family, and term
Relevance is not risk. For each relevant reference, Evalify assesses legal status (in force, lapsed, expired, abandoned, or pending), surfaces patent family context and current assignee, and estimates term/expiry — so an expired or abandoned reference is weighted very differently from an enforceable one, and pending applications are treated as a distinct future-risk category.
An application publication is resolved to its granted family sibling and the granted claims are charted; where no grant exists, it is shown as a monitored pending application and never drives a Blocked verdict.
6 · Risk assessment and FTO score
Evalify weighs the charted overlaps together with legal status to produce a risk assessment per aspect and reference, an overall FTO score across the subject matter, and a written FTO assessment. The strategy lens applied to the analysis (e.g. a clearance-oriented read) is reflected in how risk is framed and weighted.
The assessment is jurisdiction-aware: each searched market gets its own verdict, computed from the same claim-chart rules. A block confined to one market — while the product is clear in others — reads as a conditional position rather than a global block, and the score is softened in proportion to how much of your scope is clear (never reaching full clearance while any searched market is blocked). See Clearance by Jurisdiction.
The score is banded to match the opinion, so the two always tell one colour story: the opinion (Clear to Operate / Conditional / Blocked) sets the colour band — green, amber, or red — and the score positions you within it. A conditional position therefore always shows amber, never a reassuring green number, and the badge and score colours stay consistent across the web app, PDF, and Word export. See Reading Your Report.
7 · Prior art and invalidity evidence
For the strongest blockers, Evalify surfaces non-patent-literature (NPL) prior art as potential invalidity evidence — a validity angle that a time-boxed clearance search frequently omits. The search runs across several open scholarly sources, chosen per patent by technical field: OpenAlex (academic literature, all fields), arXiv (software / AI / computing preprints), PubMed (biomedical and life sciences), and IETF RFCs (networking and internet standards). Every candidate is filtered to references that genuinely predate the patent's filing and screened for relevance against the claim, so what surfaces is real prior art rather than keyword noise. Each cited reference carries its source, publication date, and a link — and the report records which sources were searched and which were actually cited.
8 · Courses of action
For each potential blocker, Evalify generates concrete mitigation strategies — design-around, licensing, challenging validity, awaiting expiry, and others — as a starting point for your strategic recommendations. It also surfaces adjacent-market opportunities where the subject matter would face more freedom.
9 · Consistency review
Before the analysis completes, Evalify runs a final consistency pass over the finished work-product to reconcile any tension between the detailed findings, the risk assessment, and the score — so the report tells one coherent story.
Where to find this in your report
Every analysis includes a Methodology section — right before Disclaimers & limitations on screen and in the PDF and Word exports — that records, for that specific matter, the scope searched (what · where · when · depth), the IPC/CPC classifications covered, and a search audit (how many references were screened, assessed, charted, and cleared). This page explains the method in general; that section shows the exact figures for your analysis, and the interactive assistant can walk you through either.
Throughout the report, an interactive assistant grounded in your specific analysis lets you interrogate any finding, reference, or term. See Reading Your Report.
Recall safeguards and adversarial review
Beyond the core stages, Counsel Grade runs additional passes that widen recall and pressure-test each clearance.
Citation-based discovery. For the strongest identified blockers, Evalify follows their forward citations — the later patents that cite them — and charts those references limitation-by-limitation like any other. It is a recall step that surfaces adjacent art an initial relevance ranking may have under-weighted. Every pulled-in reference is charted objectively; an off-target citation simply charts clear, and recall never manufactures a risk. These references carry a Forward citation provenance badge.
Competitor / assignee forced search. You can force Evalify to retrieve and chart a named competitor's on-topic patents regardless of the semantic-relevance ranking, via Expand to competitor. It intersects the competitor's ownership with the aspect's technical scope — only on-topic art, not the whole portfolio — and charts the forced references objectively. Forcing retrieval is not forcing a finding.
Prior art cited against the patent. For confirmed blockers, Evalify retrieves the references the patent office cited against that patent during its own prosecution — its backward citations, including examiner-classified references: category X (relevant to novelty on its own), Y (relevant to obviousness in combination), and A (technological background) — plus the examiner-cited literature. This is presented as invalidity evidence and grounds the Challenge Validity course of action in the actual references. It is strictly advisory and never changes the FTO score or verdict.
Adversarial claim construction. Evalify stress-tests thin Clear verdicts by reading each claim the broadest way a patent owner's litigator would assert it — broadest-reasonable construction plus the doctrine of equivalents. Where a clearance rests on a single absent limitation and a high-threat exposure credibly attacks that last barrier, the verdict is pulled from Clear to Conditional.
Independent second-AI review. A separate, independent AI review plays a plaintiff's litigator and stress-tests every clearance. It is review-only — it can never contradict the element-by-element claim chart — and material disagreements surface as review-only, threat-rated exposures.