NIW Expert Letters: What AAO Actually Counts

A good expert letter does a job; a bad one is dead weight

Expert letters are one of the few pieces of evidence USCIS quotes back at you. In the AAO decisions we have reviewed, letters show up two ways: as the language an officer leans on to grant a hard case, and as the thing flagged “not probative” in a weak one. A letter is not a character reference. It is supposed to do specific work under the Dhanasar prongs, and a letter that does not do that work is filler.

We do not have a structured count of letters per petition in our data, so this is not a numbers post. It is about what AAO has actually treated as persuasive, what gets discounted, and how to brief the people writing on your behalf.

Independent versus non-independent

Independent letters

An independent recommender has no current or past employment, supervisory, mentorship, or co-authorship tie to you. That independence is the entire point. An independent letter is proof that the wider field — not just the people who already know you — considers the work important.

The test is concrete. Published a paper together? Not independent. Sat on your dissertation committee, served as your postdoc PI, chaired your thesis defense? Not independent. Cited your work but never collaborated with you, never employed you, never advised you? That person likely qualifies, and that person’s letter is worth more than a glowing one from your own department.

Non-independent letters

Letters from supervisors, current employers, collaborators, and former advisors still have a use. They establish your day-to-day role and supply firsthand detail no outsider could give. But on Prong 2 — whether you are well positioned to advance the endeavor — AAO weighs them less. They read as employment references, because that is what they are. A record built only from people inside your circle is a thin record, no matter how many letters it contains.

What AAO faults in letters

The denial language across the corpus keeps returning to the same four problems:

1. Conclusory praise. “Dr. X is a world-class researcher whose work is foundational.” AAO calls this a conclusion with no facts behind it. The letter has to explain how the work matters, with specifics, not assert that it does.
2. Boilerplate. When several letters describe the same work in near-identical wording, an officer reasonably infers the petitioner drafted them. Identical phrasing across letters undermines the independence the letters were supposed to prove.
3. Unexplained jargon. The officer reading the file is not a specialist in your sub-field. A letter that is unreadable to a non-expert gets discounted. Strong letters define the field-internal term in one sentence, then make the field-internal point.
4. Significance with no external marker. “The most-cited paper in the sub-field” needs the citation count attached. “Published in a top journal” needs the acceptance rate or impact factor. A claim with no number behind it reads as advocacy.

How many letters? There is no magic number

There is no regulatory minimum, and our data has no “letter count” field, so anyone quoting you a precise number is guessing. What the prong structure actually tells you is more useful than a count:

• Independence matters more than volume. A stack of letters all from collaborators is weaker than a smaller set with genuine outsiders in it. Past a point, more letters dilute rather than strengthen — an officer reads the first few closely and skims the rest.
• Cover the claims you are making. If you argue national importance for cybersecurity work, you want an independent cybersecurity expert addressing exactly that. Match recommenders to the specific arguments in the petition.
• Do not orphan Prong 3. Most letters pile onto Prong 1 and ignore the waiver question entirely. At least one letter should speak to why waiving the job offer benefits the United States. The under-attended prong is where the petition is exposed.

Who to ask

Academic and research petitions

Senior faculty at institutions you have not worked with, study-section members, federal-lab PIs, journal editorial-board members in your field. The strongest letters come from people who could plausibly have peer-reviewed work in your sub-field but have never personally collaborated with you.

Industry petitions

Principal or chief engineers at firms other than your employer, research scientists at federally funded labs, standards-committee chairs. AAO tends to view a vendor selling into your industry more cautiously than a neutral party — a CISA architect writing for a network-security petitioner carries more weight than a Cisco engineer writing for the same person.

Clinical petitions

Department chairs and division heads at hospitals other than your own, specialty-society leadership, authors of relevant clinical guidelines. Current attending colleagues are easy to recruit but read as non-independent.

What a strong letter looks like, section by section

The letters that hold up tend to share a structure. Nothing requires this exact order, but it covers what an officer is looking for:

1. Identification. Name, title, institution, and the basis of the recommender’s authority. Three or four sentences.
2. Independence statement. Explicit: “I have no current or past employment, supervisory, or co-author relationship with [petitioner].” Leave it out and AAO assumes the worse reading.
3. How they came across your work. “I first encountered Dr. X’s work when I cited their 2023 paper in my own research.” Concrete, not “I have followed their career for years.”
4. The contribution, with field-level context. Two paragraphs. What the technical contribution is, and why the field cares.
5. The prong tie-in. One paragraph per relevant prong, in actual Dhanasar terms — national importance, well positioned, waiver justified.
6. Sign-off with CV. An attached CV establishes that the recommender is who the letter says they are.

Briefing recommenders without writing the letter for them

The approach that fails is the obvious one: you draft the letter, the recommender signs it. AAO is good at catching this, and the boilerplate problem above is mostly this in disguise. The approach that works:

• Send your CV, your three to five strongest contributions, and a one-paragraph summary of the proposed endeavor.
• Send a single-page brief on what the letter needs to address — which prong, the independence statement, the structural points above.
• Let them write in their own voice. Variation between letters is evidence of independence. Uniformity is evidence you wrote them.

Start earlier than feels necessary

Senior recommenders are slow. Plan eight to twelve weeks from first ask to final letter. Build the list before the petition draft is final, send asks early, and line up backups for the non-responses that will happen.

The one thing to do today

If you are assembling letters now, audit your shortlist for independence before you do anything else. Open your CV next to the list and check each name against your publication, employment, and mentorship history. Anyone who appears in both is not independent — replace them before you send a single brief.

Our $10 case review includes a per-prong evidence-gap read that flags whether your letters actually cover all three prongs, and the petition builder generates a per-letter brief for each recommender based on the prong they should address.