Strong records get denied. The file that persuades is organised around the criteria, not around the career.

Evidence · 10 min

The six questions a record has to answer. Every claim on this page is sourced in the list at the foot of it.
The question the officer asks	What actually answers it
Does this evidence objectively meet the criterion as the regulation describes it?	A primary document in the exhibit set that matches the criterion's own words: the award certificate, the letter requesting the review and proof the review was completed, the published article itself.
Is the person one of the small percentage who have risen to the top of the field?	Not one document. The pattern across the whole record, read together at the second step.
Is the claim corroborated by someone with standing to assess it?	A letter from a person outside the applicant's own institution that states the writer's basis of knowledge, not a letter that praises without explaining.
Can this be checked without taking the petitioner's word for it?	Third-party documentation: the journal's record, the conference programme, the citation data, the contract, the licence, the adopter's own publication.
Is the acclaim sustained rather than historical?	Evidence dated across years, with the most recent work inside the same area of expertise being claimed.
Is there anything the record contradicts?	Nothing, because the file presents one coherent account of the same work rather than three overlapping ones.

The short version

Two people apply for the same classification in the same quarter. One has a stronger career by every conventional measure: more citations, a better institution, a longer list of publications. The other is denied less often. The difference is almost never talent. It is that the second file is organised around the criteria and the first is organised around a career.

Those are two different shapes of document. A career is a chronology: education, then roles, then outputs, in the order they happened. A criterion is a specification: a sentence that describes one kind of evidence, with conditions attached. A file in chronological order makes the officer perform the translation from career to criterion, for every criterion, at speed, with a decision to write at the end of it. A file in criteria order performs that translation once, for the officer, and hands over the exhibit numbers.

USCIS is explicit about the two-step structure on both the O-1A and EB-1A side. Step one assesses, by a preponderance of the evidence, which evidence objectively meets the parameters of the regulatory description that applies to that type of evidence. Step two is the final merits determination, where all the evidence is evaluated together, in the context of the high level of expertise the classification requires. The first step is about fit to a specification. The second is about the standard. A file that is well organised for step one can still lose at step two, and a file that is vague at step one rarely reaches step two intact.

The organising principle follows from that. Every sentence of the petition letter should be doing one of two jobs: matching evidence to a criterion, or building the whole-record argument that the person is at the top of the field. Anything that does neither, however impressive, is a page the officer has to read without being able to use it.

This is the part that is within the applicant's control. Citation counts are what they are. The structure of the argument, the numbering of the exhibits, and the quality of the corroboration are all decisions, and they are made before filing.

The criteria are a specification, not a checklist

Read the criterion's own words before you decide whether you meet it

EB-1A criterion five is original contributions of major significance in the field. Criterion one is lesser nationally or internationally recognised prizes or awards for excellence. Criterion three is published material in professional or major trade publications or other major media, relating to the person's work. Each of those sentences contains nouns that narrow it: a prize, not a grant; about the person's work, not about the employer's; in the field, not in the person's department. A file built from a mental summary of the criteria will keep failing on words the applicant never read.

The specification has quality conditions inside it

USCIS states that for the purposes of the first step, officers should consider the quality and caliber of the evidence to determine whether a particular regulatory criterion has been met. That single sentence is the reason a case built on formal compliance fails. A peer review performed once is not automatically a judging credential; the guidance describes peer reviewing for a scholarly journal as evidenced by a request from the journal to do the review, accompanied by proof that the review was actually completed. Both halves are part of the criterion in practice.

Some criteria exclude the most obvious evidence

The published-material criterion is about the person and their work. Material about the person's employer or another organisation, and its work, is not material about the applicant. Marketing material created to sell the person's products or to promote their services is not generally considered published material about the person, and that includes seemingly objective content in major outlets. A specification read properly tells you that the glowing trade-press piece you were about to lead with is a liability, not an asset.

Being funded, patented or published is not the same as being significant

On the original-contribution criterion, the guidance states that being funded, patented or published does not by itself establish the significance of the work. A published paper proves there is a paper. A granted patent proves there is a patent. Significance has to come from somewhere else: published materials about the significance of the work, testimonials and letters about it, documentation that it has been cited at a level indicative of major significance, or patents and licences deriving from the work with evidence of commercial use. A pending patent, on its own, is generally treated as needing further supporting evidence of originality, such as detailed reference letters.

Mapping the career to the criteria is the actual work

Mapping means, for each criterion you intend to claim, naming the specific achievements that satisfy it and the specific exhibits that prove those achievements. Not the category of achievement. The achievement.

Take a product engineer with a decade in machine learning systems. Under a career-first approach, the file says: led the ranking team at a mid-size company for four years, published nine papers, spoke at five conferences, holds three patents. Under a criteria-first approach, the same facts become: judging (programme committee for two named conferences in two named years, with the invitation letters and the committee listings), original contributions of major significance (one ranking method adopted by two other named companies, evidenced by their own engineering publications and by the adopter's internal announcement), scholarly authorship (the nine papers, with each journal's international circulation established from the journal itself), and leading or critical role (the role at an organisation with a distinguished reputation, evidenced by revenue and headcount data and a letter from the person who held the budget).

The second version is not a richer story. It is the same facts, sorted so that each one lands on a criterion that already exists in the regulation. That is the whole technique, and it is mechanical enough to do deliberately.

Three rules make the mapping honest. First, one achievement can support one criterion well or several criteria badly. Reusing the same conference talk as judging, as press, and as a leading role produces a record that looks thin at step two even when the underlying career is not. Second, if an achievement does not fit any criterion, it still belongs in the record, because at the second step officers consider any potentially relevant evidence in the record, even evidence that does not fit one of the regulatory criteria. Third, count the criteria before you write. If the mapping produces two solid criteria and four stretches, the file is not an EB-1A file yet, whatever the CV says.

The mapping should be done in a table before a single line of the petition letter is drafted: criterion, achievement, exhibit number, one sentence of significance. If that table cannot be filled in, the letter cannot be written, and filing without it is how technically compliant petitions get denied.

Technically responsive evidence fails at the final merits step

Technically responsive means the evidence answers the criterion's form without establishing anything about the person's standing. It is the most common failure mode in self-prepared filings, because it is invisible from inside the applicant's own head.

The clearest example is volume. Thirty papers with high citation counts answer the authorship criterion and answer it well. They do not, by themselves, establish that the person is one of the small percentage who have risen to the very top of the field. The criteria are a threshold. The standard is a sentence about the entire field.

USCIS draws the line explicitly at the second step: officers consider all the evidence in the totality, and some evidence may weigh more favourably on its own while other evidence is more persuasive when viewed with other evidence. That instruction cuts both ways. A weak exhibit can be pulled up by context, and a strong exhibit can be isolated by a record that gives it no company. A single prestigious award surrounded by unrelated compliance evidence reads as exactly that.

The practical corollary is that the second step needs a section of the file devoted to it. Not a summary at the end, but an argument: here is the field, here is where the boundary of the top of the field sits, here is the independent evidence that places this person inside it, and here is why the totality supports the conclusion. Files that walk criteria one through three and then stop have made the officer do the hardest part of the job unassisted.

There is one instruction that protects a well-mapped file here. An officer may not deny merely because a particular type of evidence is absent if other evidence meets the regulatory criteria. That is the reason a record built criteria by criteria, with different exhibits carrying different claims, is more resilient than a record resting on one signature achievement.

Independent corroboration is what makes a claim checkable

A letter is evidence about the writer as much as the applicant

The guidance on the original-contribution criterion says detailed letters from experts in the field explaining the nature and significance of the person's contribution may provide valuable context, particularly when the record includes documentation corroborating the claimed significance. It then describes what the letters should do: specifically describe the person's contribution and its significance to the field, and set forth the basis of the writer's knowledge and expertise. A letter that does not state the basis of the writer's knowledge fails that description however senior the writer is.

Independence is a factual question, not a formality

A letter from a current supervisor is an employment reference. A letter from a co-author is a collaborator's account. Neither is independent evidence of standing, and a file composed entirely of them has no external check on it. Independent means someone who has never worked with the person, who knows the work from the field, and who can say where this work sits relative to the rest of it. One such letter is worth more than five warm ones, because it is the only kind that adds information the officer did not already have.

Corroboration means a second document, not a second assertion

The claim that a method was adopted by others is corroborated by the adopter's own publication, repository, production documentation or announcement, not by a letter asserting adoption. The claim that an award is competitive is corroborated by the awarding body's own selection numbers or criteria. The claim that a journal has international circulation is corroborated by the journal's own description of its circulation. Every one of those is a document that exists independently of the person asking for the visa, which is precisely why it counts.

Internal consistency is part of corroboration

Dates, job titles, employer names and role descriptions have to match across the petition letter, the letters of support, the form, and the exhibits. A record that describes the same project three ways across four documents is not a record with a strong case; it is a record that has given the officer a reason to read the rest of it sceptically. Consistency is cheap to check before filing and impossible to repair afterwards, because a correction after an RFE draws attention to the original inconsistency.

What an adjudicator can actually verify

The useful mental model is that the officer is trying to write a decision that survives review. That means the officer can use what is in the file, and can check what is publicly checkable, and can use what the file itself demonstrates about the record's organisation. Anything that requires the officer to take a claim on trust is weaker than a claim that arrives with its own proof.

Four things are verifiable in practice. Documents with an independent existence: the award, the journal issue, the conference programme, the contract, the diploma, the licence. Institutional records: a journal's record that a review was requested and completed, a committee roster, an employer's confirmation of title and dates. Public data: citation records, patent records, regulatory approvals, funding awards, and adoption visible in other people's published work. And the internal evidence of the file itself, which is the one people forget: an exhibit index in which every numbered exhibit is actually referenced in the letter, and letters in which every factual assertion is traceable to an exhibit, is itself evidence that the record was built deliberately.

What does not verify is assertion without a document, praise without a stated basis, and any claim whose only support is the applicant's own account. Those are not disqualifying. They are simply load-bearing only if nothing else in the record carries weight.

This is why an exhibit index does more work than it appears to. It compresses the officer's job into following references. If the letter says the person judged the work of others, and cites exhibits 7 through 14, and exhibit 7 is the journal's request and exhibit 8 is the confirmation that the review was completed, the officer can verify the criterion in two documents rather than reconstructing it from a narrative.

Two researchers, one number

This is an illustration of the pattern, not a report of one case. Two early-career researchers, both with roughly nine hundred citations and an h-index of fourteen, both claiming extraordinary ability, both filing without an employer petition.

The first file is organised around the career. It opens with a two-page biography. It lists all nineteen publications in reverse chronological order with citation counts. It includes six letters: the doctoral supervisor, two co-authors, a department head, a collaborator at another university, and one more co-author. It claims judging, authorship and original contributions, in that order, with no exhibit numbers in the letter and a bundle of PDFs at the end.

The second file is organised around the criteria. The letter opens with the standard, then takes the criteria in the order they will be examined. Judging: four conferences and two journals, each with the invitation, and for each journal review the confirmation that it was completed. Original contributions: two findings, each with the paper, then the third-party documentation of adoption — for the first, two groups that cite it and build on it in their own published work; for the second, a method incorporated into an organisation's published guidance. Authorship: the relevant papers, with each journal's circulation established from the journal's own description, and the citation record presented as a distribution rather than a total. Leading role: one project, with the funding body's own award announcement and a letter from the programme's sponsor. Then a section on the standard itself, defined against the field and evidenced by the letters.

The letters in the second file number four: three independent experts who have never co-authored with the applicant and who each explain how they know the field and where this work sits in it, and one collaborator whose letter is included for a specific factual claim only. Each letter is referenced by exhibit number from the section it supports.

The second file is not a better career. It is the same career, with each part of it placed where the officer needs to find it, and the corroboration of every contested claim attached to the claim. When the decision gets written, the first file has to be argued for by the officer. The second has already made the argument, and the officer's remaining work is to check it. That is the whole advantage, and it is available to anyone who does the mapping before writing a word.

Questions this raises

What does evidence architecture mean in an immigration petition?

It means organising the record around the regulatory criteria rather than around the person's career history. Each claimed criterion is matched to specific achievements, each achievement to specific numbered exhibits, and each contested claim to an independent document that corroborates it. The petition letter then follows the criteria, and the officer's job becomes checking references.

Why do strong careers get denied when weaker ones pass?

Because the first step asks which evidence objectively meets each criterion's description, and the second asks whether the whole record shows the required standard. A strong career presented as a chronology leaves both steps to the officer. A modest career mapped criteria by criteria can satisfy step one document by document and build a defensible second step.

How many recommendation letters do I need?

There is no number. What matters is what the letters do. USCIS expects letters that specifically describe the contribution and its significance and set out the basis of the writer's knowledge and expertise. A few independent letters that do that, tied to exhibits, outperform a larger set of warm letters from supervisors and co-authors.

Does USCIS verify what I submit?

The record is assessed by a preponderance of the evidence, which means the officer weighs what is in the file. Claims supported by third-party documents — journal records, programme rosters, citation data, patents, contracts, adopters' own publications — can be checked and carry more weight than assertions that rest only on the applicant's own account.

Can I use the same achievement for several criteria?

You can, but it usually weakens the record. One achievement used in three criteria produces three thin entries rather than one strong one, and at the final merits step the officer reads all the evidence in the totality. Reuse a single achievement only where it genuinely satisfies each criterion on its own terms.

What is the final merits determination?

It is the second step of the analysis. After step one decides which evidence objectively meets the regulatory criteria, step two evaluates all the evidence together in the context of the high level of expertise the classification requires. Passing the criteria is a threshold, not eligibility, and a case can fail there.

Sources

What it costs

Every fee we charge is published.

One number per route with the government's charge beside it, and the same number applies whether you ask us on this page or on a call.

See the fees Book a call

Keep reading

The Brief

The longer version of this argument

ANALYSIS Country of birth is the quiet variable in every timeline you have been given Two identical records, two passports, timelines years apart. ARTICLE The founder's timeline: year one, two and three The evidence has to come from real work done in the right order. A sequence, not a checklist.