USCIS approved just 32.2% of EB-1A petitions in Q3 of fiscal year 2026, the lowest rate on record after six straight quarters of decline. That means roughly two out of every three people who filed this quarter got a denial or an RFE they had to fight through. According to data tracked by Lawfully, Request for Evidence rates climbed to the 46 to 49 percent range across EB-1A and EB-2 NIW filings earlier this year.
I’ve spent 11 years writing petition letters and recommendation letters for EB-1A and EB-2 NIW self-petitioners. I am not an attorney, and nothing here is legal advice. I write the narrative that presents your evidence to an adjudicator. That’s it. But after more than a decade of doing that work, I want to show you something most guides won’t: the actual language USCIS uses when it decides a petition falls short, pulled from two real RFEs, with every identifying detail removed. Statistics might not reveal everything, but anecdotal evidence does. Lets understand USCIS based on different markers that make up an EB1A petition.
RFE one: the activity versus proof problem
The first RFE I’m quoting from addressed four criteria for a petitioner in tech. USCIS accepted one of them outright.
On the judging criterion, the officer wrote that the petitioner had “not shown that your participation in projects and providing and soliciting feedback in an informal capacity equates to participation as a judge of the work of others in the field.” The petitioner had submitted an email accepting an invitation to review. USCIS called that “no probative value since it does not establish your completion of a review, only presupposes it.”
An invitation to review is not evidence of review. Only proof the review was actually completed and submitted counts.
On the leading or critical role criterion, the letter was equally specific: “A leading role should be apparent by its position in the overall hierarchy of an organization… a critical role is evidenced by its overall influence on an organization or establishment as a whole.” The petitioner had submitted letters describing good work on various projects. USCIS’s response: those letters “does not contain detailed and probative information that specifically addresses how your role was leading or critical for an entire organization or establishment.”
On high salary, the officer rejected a comparison against “the base salary of other tech professionals,” insisting the comparison needed to be against “other Systems Software Developers” specifically, and noting that Department of Labor wage data alone, without corroborating evidence, doesn’t establish a significantly high salary.
The one criterion this petitioner cleared, original contributions of major significance, got a single line in response: “The evidence meets the plain language of the criterion.” No pushback. That’s what it looks like when a petition is specific enough that there’s nothing left to question.
RFE two: five criteria, five different failure modes, and an AI flag
The second RFE I read is more detailed, and one line in it should get every self-petitioner’s attention in 2026.
On association membership, the petitioner listed memberships in several recognized professional bodies. USCIS didn’t dispute the memberships. It rejected the evidence because she “did not submit any documentary evidence such as association bylaws which demonstrate that outstanding achievements are required for membership, and that eligibility is judged by recognized national or international experts in their field.” Belonging to a well-known organization proves nothing on its own. The bylaws proving selectivity are the actual evidence.
On published material about her work, she submitted articles from several outlets, including one major national publication. USCIS accepted that the publication itself qualified as major media, then rejected the article anyway: it “does not identify the author, and it is not about the petitioner and her work in the field, but rather it only quotes her about her opinion on the subject matter.” Two other articles were flagged as likely promotional placements, since the RFE cited USCIS Policy Manual language directly: “Marketing materials created for the purpose of selling the person’s products or promoting the person’s services are not generally considered to be published material about the person.” Being quoted is not the same as being profiled. And placed content reads as placed content to an adjudicator trained to spot it.
On judging, the pattern from the first RFE repeated almost exactly. She submitted emails and a press release describing her participation as a judge for two award programs and reviewer work on technology proposals. USCIS found she’d judged “the work of companies and projects” but never submitted evidence she’d judged “the work of other software developers or of others in an allied field.” The criterion requires judging peers in your own field, not evaluating submissions in a general sense.
On original contributions of major significance, she submitted a patent for an AI-based medical detection apparatus. USCIS didn’t dispute the patent’s existence, but noted that “the issuance of a patent, by itself, does not verify the significance of the innovation because the patent’s significance is not evaluated during the application process.” No evidence was submitted showing the patent had been licensed, generated commercial activity, or been adopted by anyone.
Then there’s the detail that should stop you if you’re tempted to speed up your petition with AI-written content. The petitioner submitted three scholarly articles published a few months before filing. USCIS flagged them directly: “There are concerns about whether these articles were generated by artificial intelligence due to their odd structures and wording, repetitive phrases, and acronyms that are repeatedly defined or that are inconsistently used.” The RFE went further, stating the articles “appear to have been made specifically for this criterion in an attempt to elevate her acclaim.”
That’s not a hypothetical risk. That’s an adjudicator, in a real case, naming AI-generated text as a reason evidence was rejected. If USCIS officers are trained to spot repetitive phrasing and inconsistently defined acronyms in scholarly articles, the same scrutiny applies to any part of your petition that reads like it was generated rather than written.
The pattern underneath both letters
Ten different rejected sub-criteria across two RFEs, and every one comes back to the same root issue. The evidence described activity, association, or output. It didn’t establish position, selectivity, authorship, or verified impact.
“I’m a member of a respected association” is association. “This association’s bylaws require outstanding achievement, judged by named international experts on its admissions panel” is selectivity.
“I was quoted in a major publication” is activity. “This publication profiled me and my work specifically, with a named author and no commercial relationship to my employer” is proof.
“I hold a patent” is output. “My patent has been licensed by these companies, cited in this many subsequent filings, and generated this documented commercial result” is verified impact.
USCIS isn’t asking whether you’re accomplished. It’s asking whether the specific documentary evidence in the file, read on its own, proves position, selectivity, authorship, or impact without requiring the officer to take anything on faith. That’s a narrower standard than most technical professionals expect when they sit down to write their own case.
What happened after the RFE response, and what that tells you about the evidence
The petitioner in the second case responded to the RFE, and the response was denied. Her EB-1A didn’t go through. That outcome is worth sitting with, because it tells you something the RFE alone doesn’t: this wasn’t a case where one weak criterion sank an otherwise strong petition. Five separate criteria had five separate documentation gaps, and a response that didn’t close all five in a way the officer found convincing led to a final denial.
If you’re weighing whether to refile after a denial like this, here’s what’s a documentation question, and what isn’t. Whether to refile, appeal, or pursue a different visa category entirely is a legal strategy decision, and that call belongs with a licensed immigration attorney, not with a petition writer. What I can tell you, from the writing and evidence side, is that refiling with the same documentation, reorganized or restated, tends to draw the same scrutiny, because the same evidentiary gaps are still sitting in the file. If an attorney advises a refile, the evidence itself needs to change, not just the language around it.
Concretely, that means association membership needs actual bylaws showing the group requires outstanding achievement and judges applicants through recognized experts, not just a certificate of membership. Published material needs a piece that profiles the work itself, with a named, independent author, not another quote pulled into someone else’s article. Judging needs documentation of having evaluated peers in the same field, not company submissions or general proposals. A patent needs proof it did something in the world: licensing, citation, adoption, commercial use. And anything resembling AI-generated writing needs to be replaced with material that reads as unmistakably human, because once USCIS has flagged that concern in a case, it’s reasonable to expect the next filing gets a closer read.
Which category to file under, and whether to refile at all, are decisions for a qualified immigration attorney who can assess the full case. What we do is make sure that whichever documentation gets submitted actually proves what it needs to prove.
Why this is getting harder, not easier, in 2026
RFE rates climbed into the 46 to 49 percent range earlier this year, and approval rates dropped to 32.2 percent in the most recent quarter. Neither RFE above asked the petitioner to manufacture new achievements. Both asked for the same achievements, proven with documentation that removes the need for the officer to infer anything.
That’s consistent with what I’ve seen across our own clients over the years. Ajay Gupta, a software engineer, came to us with 50-50 odds according to attorneys he’d consulted, and no advanced degree. His EB-1A was approved in five months once his case built position-and-impact proof rather than describing his work in general terms. Mahesh Patil, an embedded engineer, had accomplishments outside academia that had never been surfaced or connected to a specific criterion. His case was approved in seven months. SR Reddy, a senior scientist, had forgotten achievements that, once verified and framed against a specific criterion, helped his case get approved in four months. Manasa N, an Indian classical dancer, was told by attorneys she had a low chance of success before her petition was built out with two months of additional evidence work. Her case was approved too.
None of these outcomes involved adding anything untrue. They involved proving what was already true in the exact way USCIS requires it proven.
What to check before you file
Go through your own draft, criterion by criterion, and ask whether your evidence proves position, selectivity, authorship, or impact, or whether it just describes activity. An invitation is not completion. A quote is not a profile. A patent is not proof of significance. Membership is not proof of selectivity. And if any part of your petition was drafted with AI assistance, have a human editor rewrite it in your own voice before it goes anywhere near your filing. USCIS is actively watching for it, and one RFE already says so in writing.
None of this replaces legal judgment about which criteria you actually qualify under, whether EB-1A or EB-2 NIW fits your case, or how to respond to a prior denial. That’s a licensed immigration attorney’s call. What my team and I do, since 2015, is the writing and evidence-organization work underneath that legal strategy: turning documented achievements into the specific, verifiable proof these two RFEs make clear USCIS is looking for.
If you want a second opinion on your evidence before you file
We offer a free writing-readiness review. Send your CV and background, and within 24 hours we’ll tell you honestly whether there’s enough documented material to build a strong narrative, and roughly what that project would involve. No obligation.
Get a free writing-readiness review
If you want a closer look at your documented evidence before deciding anything, I also run a paid 45-minute evidence review where we go through your achievements criterion by criterion and flag exactly where the documentation needs to be stronger.
Book a 45-minute evidence review
We write your petition. You stay in control of every word, and you file it yourself. For legal advice on eligibility, category selection, refiling, or filing strategy, talk to a licensed immigration attorney.
Results vary based on individual qualifications, evidence quality, and adjudication by USCIS. RFE language quoted above is from two actual cases, with all identifying details removed. Client outcomes described reflect specific past cases and are not a guarantee of any outcome for future applicants.