The Kazarian two-step: how to document a final merits determination that survives review

Kazarian v. USCIS established a review structure that has now been USCIS policy for well over a decade. At step one, the adjudicator counts: does the record contain evidence satisfying at least three of the ten regulatory criteria at 8 CFR 204.5(h)(3). At step two, the adjudicator asks a qualitative question about the whole record — does it establish sustained national or international acclaim placing the beneficiary among the small percentage at the very top of the field.

The two questions are genuinely different, and the answer to the first does not determine the answer to the second. A petition can satisfy five criteria and still fail the final merits determination. That happens more often than practitioners expect, and it is nearly always in petitions that treated step one as the entire analysis.

Petition packets are usually organised by criterion, because that is how the evidence is gathered and how the regulation is written. Exhibit tabs run Awards, Memberships, Published Material, Judging, Original Contributions, and so on. The cover letter follows the same order, argues each criterion in turn, and then concludes.

That structure quietly omits step two. There is no section that steps back and argues the totality. When there is no such section, the adjudicator constructs the final merits analysis themselves from the material in front of them, which means the petitioner has surrendered the framing of the most subjective judgement in the process.

A final merits section should do three things the criterion-by-criterion analysis cannot. First, it should establish a comparative frame: what does the top of this field look like, and by what measure. Second, it should place the beneficiary within that frame using evidence already in the record, citing exhibit page ranges rather than restating them. Third, it should address the counter-reading — the strongest version of why an adjudicator might not be persuaded — and answer it.

Comparative eminence evidence is what most records lack. That can be field-level citation baselines, the size of the population holding a given membership, the acceptance rate of a venue, or the number of practitioners performing a named critical role. It is rarely in the client's document dump and almost always has to be constructed by the firm.

• Open with the comparative frame, not with a summary of criteria met.
• Cite exhibit page ranges rather than re-describing the evidence.
• Name the field precisely — 'computational materials science' beats 'engineering'.
• Address the strongest counter-reading explicitly rather than hoping it is not raised.

Arguing the same exhibit twice — once under a criterion, once in final merits — only works if the citations resolve. If Exhibit 5.3 moves because an exhibit was added upstream, every citation to it in both sections has to move with it. In manual assembly that is a find-and-replace exercise performed under deadline, and it is where broken citations enter the record.

Continuous Bates numbering across the binder with citations generated against the assembled range removes the failure mode entirely. Reorder an exhibit and every reference in both sections renumbers together. It is a mechanical fix to what looks like a drafting problem.