Screening determines which candidates receive deeper analysis. If its reasons are not recorded, a later reviewer cannot see why some documents were retained and others were set aside.
Key takeaways
- Screening applies the defined FTO scope to the candidate set. Excluded candidates should remain traceable in the record, even though they do not proceed to detailed charting.
- Record why each candidate was excluded, not just which candidates were excluded. A list of what survived is not a record; a list of why the others did not is.
- Keep an explicit route for unresolved candidates. A binary label is not enough unless the record separately preserves uncertainty and the action needed.
Screening is where the analysis is decided
Everything downstream inherits this step. A claim chart can be flawless and still be about the wrong patents. A conclusion can be carefully scoped and still rest on a candidate set that dropped the one document that mattered.
When AI-assisted retrieval produces a large set quickly, screening can become a bottleneck. Decide how the set will be reviewed and what will remain unreviewed, rather than letting the speed or ranking of retrieval set the standard.
The response is not to screen more slowly. It is to screen against stated criteria and leave a record.
Four questions per candidate
1. Could the claims be relevant to the defined product or activities? Use the title, abstract, classifications, and claim text to orient the review. Do not exclude a candidate merely because it uses different terminology, describes another application, or addresses a different problem. A claim may still warrant comparison with the product.
2. What is the status, and where? Check relevant family members separately for the places where the planned activities occur, not only where the product will be sold. Record the status source and check date. Family members can have different claims and status; an apparent lapse may require checking restoration or other unresolved events before exclusion.
3. Do the claims warrant element-level comparison? Review the potentially relevant independent claims and their limitations, not just the first claim or the title. Where exclusion depends on an element being absent, identify the product evidence and the interpretation supporting that finding. Check other relevant claims and family members before excluding the candidate; unresolved construction or equivalents issues need professional review.
4. Is the available evidence enough to exclude it? Establish what has been checked and what the available records support. A quick screen may be well founded; a slow one may still rely on an unsupported assumption. Record unresolved evidence gaps instead of treating the abstract as a complete basis for exclusion.
Four working labels
The four labels below are one way to separate immediate claim review, reasoned exclusion, monitoring, and unresolved questions. Adapt them to the workflow, but do not let a status label hide missing technical or product evidence.
| Label | Means | What it commits you to |
|---|---|---|
| Include for claim review | The candidate warrants element-level comparison | Assign it for detailed review |
| Exclude, with reason | The recorded evidence supports setting it aside within the defined scope | Retain the basis and reconsider if relevant facts, claims, status, or scope change |
| Monitor | A potentially relevant development needs to be watched | Name the trigger, owner, and review date |
| Unresolved | Material status, claim-text, product-evidence, or mapping questions remain | Assign follow-up; resolve the issue or state how it limits the assessment |
Monitor a pending application when its development could matter to the defined activities. Pending status alone does not mean every application must be monitored, and it does not by itself justify exclusion. Record why the application matters and what event would prompt review. Treat unverified status or insufficient product evidence as unresolved, not as an exclusion reason.
None of these labels is an infringement conclusion. They organise work.
Build the candidate set for claim review. Start from your product description in FTO Search, then check the candidates against your screening criteria.
What to write down
Keep the three decision fields below alongside the candidate identifier, jurisdiction, claim version reviewed, status source and date, and reviewer. These fields describe the screening decision; they are not the entire audit record.
The label. One of the four above.
The reason, in one line. For example: "Claim 1 requires a mechanical linkage; specification X, revision Y, and drawing Z support its absence in the assessed configuration." State which other potentially relevant claims were checked. "Not relevant" is a label, not an explanation.
What would change it. Name the event or missing evidence that would prompt another look. For monitored or unresolved candidates, assign an owner and a review date. For exclusions, retain any material condition on which the decision depends.
The record should let a later reviewer understand why each candidate entered or left detailed review. Missing reasons make the screening difficult to audit; they do not prove that no review took place.
Two ways screening goes wrong
Screening on relevance alone. Technical similarity is an aid to finding candidates, not the legal test. A superficially distant document may still contain relevant claims, while a close document may have no relevant right in the jurisdictions and activities being assessed. Keep technical relevance, status, and geographic scope distinct.
Letting the ranking do the screening. A low rank is not an exclusion reason. Where the review is limited, record the stopping rule, the portion reviewed, and the portion left unreviewed. Consider checks of lower-ranked results where needed. A stopping rule documents a limit; it does not prove that the unreviewed material is irrelevant.
How preliminary screening appears in Eureka
The Eureka FTO Search workflow starts from a product description. The description is broken into technical features, candidate patents are screened, and selected claims are compared with the product features. The resulting comparisons give reviewers a structured starting point for checking the claim language, proposed correspondence, and supporting evidence.
At the screening stage, check candidates against the defined product, activities, and markets before proceeding to detailed comparisons. Verify material claim and status information, and retain the reasons for exclusions and unresolved questions in your review record. The four working labels in this guide are suggested review categories, not Eureka interface labels; a generated ranking does not replace screening criteria.
This article is general information about analysis method. It is not legal advice. Screening labels organise preliminary evidence and are not infringement determinations. Claim scope, legal status, and infringement depend on the applicable jurisdiction and facts, and require qualified professional advice.
