Why USCIS Discounts Template Recommendation Letters

Why USCIS Discounts Template Recommendation Letters (and How to Tell If Yours Reads Like One)

The government told everyone, in writing, which letters it ignores. Most petition letter sets are still written exactly that way. Here are the agency's own words — and six tests to run on your drafts before anyone signs.

Every O-1, EB-1A, and EB-2 NIW petition arrives at USCIS with a stack of recommendation letters. Adjudicators read thousands of them. And the agency has been unusually candid about what it does with the generic ones: it discounts them.

This post collects the receipts — the USCIS Policy Manual's own language, and the practitioner warnings about lookalike letter sets — and turns them into a self-diagnosis checklist. It is part of our immigration letter series; the complete guide covers the genre end to end.

USCIS's own words

Three statements, straight from the Policy Manual (Vol. 6, Part F, Chapter 2) and its September 2023 and January 2025 updates (summarized in the agency's guidance alert and by firms like Ogletree):

1. Letters that "merely reiterate USCIS's definitions" of the classification are "generally not persuasive." If a letter says the beneficiary has "extraordinary ability" and "sustained national or international acclaim," it is quoting the regulation back at the person who enforces it.
2. Letters that make "general and expansive statements" about the beneficiary and their accomplishments are, likewise, generally not persuasive. "One of the leading experts of his generation" is a general and expansive statement.
3. Letters "should not form the cornerstone" of the case — testimonial claims must be corroborated by independent, verifiable evidence elsewhere in the record. An uncorroborated superlative doesn't just add nothing; it advertises that nothing backs it.

There is a fourth pattern the regulations don't spell out but practitioners warn about consistently: letter sets that read as one document. When five "different" experts submit letters with the same structure, the same pet phrases, and the same rhythm, adjudicators notice — firms advising on extraordinary-ability cases explicitly counsel varying style, structure, and voice across the set (EB1A Experts; Thomas Vallen).

Why template letters happen (it's structural, not laziness)

The template problem is baked into how these letters get made:

• One drafter, many signatures. As covered in the complete guide, the open industry norm is that the attorney or self-petitioner drafts the letters and the experts review, edit, and sign (Chen/WeGreened describes its drafting workflow publicly; Manifest Law calls attorney-prepared drafts common). One tired author writing letters five through eight at midnight produces one voice, five times.
• Everyone starts from the same samples. The top-ranking "sample letter" pages are a handful of law-firm template libraries. Thousands of petitioners fill in the same blanks and keep the same connective tissue — "it is my distinct pleasure," "I am confident that," "contributions of major significance."
• Increasingly: the same AI, prompted the same way. Generic prompts to a chatbot produce statistically similar prose. Admissions offices already face this exact blind spot with academic recommendation letters — none of the 174 university AI policies we analyzed even mentions LORs, while teachers openly use chatbots to draft them. Petition letters have the same exposure with higher stakes: the signer is attesting, over a signature block on letterhead, to prose no human wrote or checked.

None of this makes drafting-for-signature wrong — it makes pre-signature quality control mandatory. The fix isn't to stop drafting; it's to catch template-shape before the expert signs.

The six tests: does your letter read like a template?

Run these on each draft, in order of severity.

1. The definition-echo test

Search the draft for the regulation's own vocabulary: "extraordinary ability," "sustained acclaim," "small percentage at the very top," "major significance," "national importance." Every hit is a sentence asserting the legal conclusion instead of evidencing it — the exact pattern USCIS names as not persuasive. Rewrite each one as the fact that would justify the phrase.

2. The swap test

Replace the beneficiary's name with someone else in the same field. Does the letter still read fine? Then it contains no evidence specific to you, and it will read that way to an officer. A strong letter fails the swap test in every paragraph — the named papers, deployments, numbers, and incidents belong to one person only.

3. The evidence-density test

Count, per page: numbers, named artifacts (papers, products, patents, standards), named organizations, and checkable events. Strong petition letters run several per paragraph. If a full page has zero, that page is praise, and praise is what the Policy Manual calls "general and expansive."

4. The corroboration test

For each factual claim, ask: does an exhibit elsewhere in the record back this up — and could the signer personally defend it? Claims the record can't corroborate violate the "cornerstone" guidance; claims the signer can't defend should never go out for signature at all (support statements are commonly executed as sworn declarations — see the accountability section of the complete guide).

5. The lay-reader test

Hand the draft to someone outside the field for three minutes. Can they say what you did and why it matters? The actual reader is a USCIS generalist with minutes per exhibit, not a peer reviewer. Jargon-dense letters aren't impressive; they're unread.

6. The one-voice test (the set-level check)

Line up all five to eight drafts. Same opening move? Same paragraph order? Same distinctive phrases ("I wholeheartedly recommend," "truly groundbreaking") recurring across "independent" authors? Same sentence rhythm? That is the lookalike-set pattern practitioners warn adjudicators discount. Each recommender's edits should be genuinely incorporated — an expert who rewrites two paragraphs in their own words is doing you a favor, not creating work.

What to do with a letter that fails

Don't decorate it — rebuild the failing paragraphs around evidence:

1. Pull the specific claims from your own record that this recommender can honestly attest to (this letter's assigned criterion or prong — see the O-1 criteria mapping and the Dhanasar prong mapping).
2. Rewrite every echo-test hit as a fact with a number or a name in it.
3. Restore the recommender's actual voice: their discipline's vocabulary, their real anecdotes, their honest hedges. A letter with one candid limitation ("I can't speak to her later industrial work, but…") reads more credible than uniform superlatives.
4. Re-run the six tests on the revised set — especially the one-voice test, which only shows up at set level.

The same discipline applies outside the US system: the UK's Global Talent route requires exactly three letters, and generic letters are reported as the top reason endorsements get refused.

Where GradPilot fits

The six tests above are exactly the kind of checks that benefit from a second, systematic reader — and the pre-signature window is the moment to run them, because after signature every fix costs a favor.

GradPilot reviews the petition letter you're drafting: AI detection (was this prose machine-generated — including sections a paid preparer may have generated without telling you, a risk we've documented with education agents and visa statements), plus structured feedback on specificity, definition-echo, criterion mapping, and lay readability. Like our visa statement review, it reviews what you wrote — we never write letters, because a letter only works if the signer genuinely owns it. Start at the immigration review page.

FAQ

Does USCIS actually reject template recommendation letters?

USCIS doesn't reject a petition for template letters alone — it discounts them. The Policy Manual states letters that merely reiterate USCIS definitions or make general and expansive statements are "generally not persuasive," and that letters can't be the cornerstone of a case without corroboration. Discounted letters mean the criteria they were supposed to evidence go unproven.

Can I start from a sample letter I found online?

Using a sample for structure (credentials → relationship → specific contributions → lay impact → endorsement) is standard. The danger is keeping the sample's language — thousands of petitions recycle the same phrases, and generic prose is exactly what officers are told to discount. Borrow the skeleton, never the sentences.

Is it a problem if all my letters were drafted by the same person?

One drafter is the industry norm — but one voice across the signed set is a documented risk, and practitioners advise deliberately varying structure and style between letters. The strongest protection is genuine recommender editing: each expert reshaping their letter in their own words before signing.

Can USCIS tell if a recommendation letter was written by AI?

USCIS has not published any policy on AI detection in petition letters. The practical risk is subtler: same-prompt AI drafts converge on similar phrasing across a letter set, compounding the lookalike problem officers already discount — and a signer attesting to machine-generated claims they never verified takes a real risk. Checking drafts with AI detection before signature is cheap insurance.

How do I fix a letter that is all praise?

Anchor every evaluative sentence to a checkable fact: a named artifact, a number, an organization, an event. Then cut whatever praise remains unanchored. A two-page letter with twelve verifiable facts beats a four-page letter with forty adjectives.

---

This article is about letter craft and quality control, not petition strategy, and is not legal advice. Verify current guidance against the USCIS Policy Manual and consult a licensed immigration attorney about your case.

Sources

• USCIS Policy Manual, Vol. 6, Part F, Chapter 2 — Extraordinary Ability
• USCIS Policy Manual update, September 2023 (PDF)
• USCIS Policy Manual update, January 2025 (PDF)
• USCIS alert — new guidance on EB-1 eligibility criteria
• Ogletree — USCIS updates Policy Manual on EB-1A evidence
• EB1A Experts — crafting high-impact recommendation letters (letter-set similarity warning)
• Thomas Vallen — EB-1A and NIW petitions (vary letters across the set)
• Chen Immigration / WeGreened — how we draft reference letters for clients
• Manifest Law — O-1 recommendation letter samples (attorney-drafting norm)