We read 35 example files - O-1 letters and templates for arts, film, dance, music, design, marketing and tech founders, EB-1 letters for a beauty professional and a set of science letters - and wrote down their structure in our own words. Names, titles and passages are not reproduced: 11 of the files name real people.
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
Who writes the strong letters
What the letter points to
Claims that carry weight
Where the examples were weak
The weak spots are measured against what officers discount in real final-merits refusals: letters from people who know you personally, identical phrasing, echoing the regulation, praise without facts, and promises about the future.
Neither the regulation nor the Policy Manual sets any requirement about typeface, wet-ink signatures, letterhead or the referee’s home address. Advice of that kind is folklore. The scaffold is built around the three things that are actually checked.
First, whether the author can be identified and verified: position, organization, professional connection to you. Second, independence: a letter from your direct manager weighs less than one from a specialist who knows your work but has no employment relationship with you. Third, specificity: "an outstanding specialist" proves nothing; what works is a described episode - what you did, what came of it, and how it affected the writer’s work.
One detail visible in refusals: five letters sharing one outline show regardless of formatting. What draws attention is not a shared typeface but a shared plan and recurring phrasing.
On number and mix. There is no fixed count, but four to six letters, half of them from people with no working relationship to you, read more strongly than ten from colleagues and managers. Each letter should carry something of its own: one referee on the contribution, another on adoption, a third on recognition in the field. Three letters about the same thing weigh as much as one.
Myth: It is mandatory to have 8-10 recommendation letters, and each one must be from someone the applicant has personally interacted with.
In reality: There is no fixed required number of recommendation letters. What matters more is their evidentiary value, independence, specificity, and relation to the claimed criteria; the letters may be from experts who know the applicant or from independent specialists, if they can reasonably speak to the applicant's contributions.
O-1 · IT - open source · outcome not stated
A developer built an independent open source project, with code written entirely by him, and wanted to confirm the significance of his work.
What they did: Received recommendation letters from organizations and specialists who used his open source work in commercial products.
O-1 · business/entrepreneurship · outcome not stated
The applicant was preparing a nonimmigrant visa through an agency arrangement and was looking for confirmations from outside organizations.
What they did: He collected several recommendation letters signed by representatives of major companies where acquaintances worked; the employers are generalized as large commercial companies.
EB-1A or EB-2 NIW · not specified · denied
In one appellate analysis of an immigration case, the focus was on the quality of the evidence, recommendation letters, and expert letters, rather than their quantity.
What they did: The applicant tried to challenge the negative decision based on the submitted letters and evidence of achievements.
Distilled from the public community chats, anonymised; corpus as of June 2026. These are participants' observations, not legal advice and not USCIS statistics.