Statistics-included
Panels judge the endeavor, not the field.The most-repeated error in the denial record is arguing that the field matters, not the specific endeavor.Open finding →The Findings Deck
What the record shows about why NIW cases turn. Open a card for the full finding; save the ones you want to come back to.
Each record describes itself: the appeal record contains only denied-then-appealed cases; sustained decisions are the wins inside that same record. Neither is the population of all petitions, and no figure here predicts an individual outcome.
4,420coded AAO appeal decisions on NIW petitions (is_niw=1) stand behind this deck.
3,997 Dismissed
91 Sustained
326 Remanded
6 Other
864 Failed the EB-2 threshold
3,562 Reached the merits
Coded AAO appeal decisions on NIW petitions (is_niw=1), 2016-2026 — appealed denials, not the filing population (D26). Petition-stage figures live in quarterly_pulse.json and never share this file.
Separately banded: 7,413 law-firm published approvals, survivor-selected twice — 4,959 NIW · 2,454 EB-1. A distinct corpus, held apart from every AAO figure (S83).
0 of 16 opened
The endeavor · Prong 1
3Statistics-included
A vague endeavor does not survive.Whether the proposed endeavor is specific enough to evaluate is the single variable the panels name.Open finding →Statistics-included
National scale must be argued, not assumed.Impact confined to clients, an employer, or a region populates the denial record.Open finding →Positioning & evidence · Prong 2
4Statistics-included
Evidence aimed at the person loses; evidence aimed at the endeavor wins.The denial record is full of credentials filed where the panel asked about the endeavor.Open finding →Statistics-included
Panels ask for traction, not degrees.A degree is the one exhibit the denial record concedes — and still finds insufficient.Open finding →Qualitative
A letter should trace what your work changed, not how good you are.A letter earns weight by naming the endeavor and the downstream work it shaped — not by the prestige of the signer.Open finding →Qualitative
An h-index screenshot argues nothing — the reason for the citations does.Bibliometrics show that work was noticed; they do not show why it mattered or that it was adopted.Open finding →The theory of the case
2Statistics-included
“My company will create jobs” is the losing theory.Unsupported economic and job-creation claims dominate the denial record; founders are the sharpest case.Open finding →Statistics-included
Who vouches matters more than how many vouch.Collaborator letters sit on both sides and separate nothing; independent validation concentrates on the win side.Open finding →How the record actually decides
4Statistics-included
The fight is won at the framing layer, not the evidence pile.The same evidence types are credited in wins and discounted in losses — the difference is posture, not the height of the stack.Open finding →Qualitative
You don’t win the balancing test — you earn it.Prong 3 almost never fails on its own; win Prongs 1 and 2 and the third follows.Open finding →Qualitative
You are being asked for “more likely than not,” not “extraordinary.”NIW is decided on preponderance of the evidence — it is not the extraordinary-ability test.Open finding →Qualitative
You are failing a consistent standard, not a moving one.The written NIW standard has been stable since Dhanasar; what tightened is how strictly it is applied.Open finding →Procedure & timing
2Qualitative
The endeavor is locked at filing — you cannot swap it later.An RFE or appeal can add evidence for the endeavor you filed; it cannot substitute a better-framed one.Open finding →Qualitative
A three-prong RFE is normal, not a verdict.RFEs question all three prongs; denials land on Prong 1 — put your weight where cases actually turn.Open finding →By profile
1Non-precedent AAO decisions bind only the parties to each case. Nothing here is legal advice or a prediction of any outcome.
AAO appeal decisions through 21 July 2026.
USCIS petition-stage data · a different population from the deck above
The I-140 record, quarter by quarter.
Official USCIS figures for every I-140 classification, FY2023Q1 to FY2026Q2. The routes a person can file for themselves sit beside the routes an employer files for them, on the same axis and the same quarters.
The gap between self-petitioned and employer-sponsored routes is in the official data. What decides any individual petition is still the prong level, which is where the record can actually help. → See which prongs decide cases
Approval rate of completions
386,312 completed cases in view · 14 quarters- NIW
- EB1-A extraordinary ability
- EB2 advanced degree (PERM)
- EB1-B outstanding prof/researcher
Approvals as a share of completed cases (approvals + denials) in the quarter. Cases still pending are not in the denominator.
- NIW
- 85.0% → 48.1%(-36.9 pp)
- EB1-A extraordinary ability
- 74.2% → 41.7%(-32.5 pp)
- EB2 advanced degree (PERM)
- 96.0% → 94.1%(-1.9 pp)
- EB1-B outstanding prof/researcher
- 92.1% → 93.8%(+1.6 pp)
What this chart does and does not establish
The divergence between self-petitioned and employer-sponsored classifications is a fact in the official data. Its cause is not. Stricter adjudication and a change in who is filing are both consistent with these numbers — NIW receipts rose sharply over the same quarters, and a demand surge plausibly changes the mix of petitions. EB2 advanced degree is a partial control, not a clean one: those cases are pre-filtered by an employer and the labor-certification process. This board shows the divergence and names the confound. It does not identify a mechanism.
Filter by profile
These are switched off, and not because the data is behind a login. USCIS publishes counts, not case records — no petition-level attributes exist in any public filing dataset. Nobody can filter official approval trends by degree, visa status, or publication record, because those fields were never collected into any public release. That is also why no honest source can quote you a personal approval probability.
Official USCIS I-140 quarterly performance data, FY2023Q1 to FY2026Q2. Counts are published by USCIS and unmodified; every rate on this board is computed from those counts.
These are official flows over whole populations. The coded record beneath them is read profile by profile, and a placement shows where yours sits. See where your profile sits — 90 seconds →