The Letter Nobody Actually Reads (But Still Gets You In)
I spent four years writing recommendation letters for people applying to grad school, fellowships, and a handful of industry roles where the committee was actually reading them. You'd be surprised how many good candidates sink because their recommender wrote a generic five-paragraph essay that says nothing anyone couldn't guess. The process of How To Write A Recommendation Letter isn't complicated, but it is undervalued by most people who've never had to produce one at a high-stakes level. Let me walk you through what actually works and where most people mess it up. Start by understanding what the reader actually cares about. Admissions committees and hiring panels want three things in order: evidence that you can do the work, some proof that you'll survive the pressure, and a signal that someone who knows you well is willing to stake their own reputation on vouching for you. Everything else is decoration. I've read letters that ran three pages of flowery prose about how hardworking and creative someone was, and they were useless because there wasn't a single concrete example anchored to a real project or outcome. The structure I use when I sit down to write one is roughly this. You open with your relationship to the candidate and how long you've known them. That's it for the first paragraph, and it should be two or three sentences max. Nobody needs to know that you took their introductory course in sophomore year and got a B-plus. They need to know whether you've supervised their actual work over time, under real conditions.
The body paragraphs are where letters usually go wrong. Each one should focus on a single quality and support it with one specific incident. Not a general description of the person's habits. An incident. When Sarah was building the logistics model for our senior project, she hit a wall on day three where the simulation kept crashing because of a memory leak she couldn't isolate. She spent six hours debugging it alone, then rewrote the data ingestion module from scratch the next morning. That's the kind of sentence that matters. The alternative reads like "Sarah is a determined and analytical thinker who overcomes challenges with perseverance," which is exactly what every other letter says. Here's the thing that most writers miss: specificity costs almost nothing and the payoff is disproportionate. A single precise detail beats two paragraphs of generic praise every time. It signals to the reader that you actually observed the person rather than filling out a form.
Common Pitfalls That Sink Letters Before They Start
The biggest mistake I see is writing about the candidate's grades instead of their work. A 4.0 GPA is already in their transcript. The letter needs to tell you something the transcript can't. I had a student once whose advisor wrote him a letter that mostly summarized his coursework and mentioned he was "always punctual and respectful." That's not a recommendation. That's a background check result. The letter should describe what the person does when things go wrong, how they handle ambiguity, how they treat peers under stress, what happens when they're given ownership of a difficult problem. Another issue is the comparison trap. Committee members want ranking information, but most recommenders stumble here. You don't need to say someone is in the top one percent of students you've ever taught. That sounds performative and unverifiable. A better approach is to ground the comparison in scope. Something like "I've supervised approximately forty undergraduates across six years of lab work, and Alex is among the top five in terms of independent problem-solving ability" carries far more weight because it's anchored to a real denominator. I also see a lot of letters that overstate qualities the writer didn't directly observe. If you taught someone in a large lecture hall with three hundred students, you can't credibly speak to their collaboration skills or how they handle group dynamics. Stick to what you've seen firsthand. Honesty about the limits of your observation actually strengthens the letter rather than weakening it.
Get the Full Details
The Structure That Actually Works In Practice
Here's a practical breakdown of what goes on the page. The opening paragraph establishes your credentials and your relationship to the candidate. One paragraph. Keep it tight. The second section addresses the candidate's technical or analytical ability with a detailed example. This is where you describe a specific project, problem, or assignment and explain what the candidate did and why it mattered. Two paragraphs here is normal. Don't compress it. The third section covers soft skills, work ethic, or interpersonal dynamics, again backed by a concrete moment. This is separate from the technical section because committees evaluate these dimensions independently. A candidate can be brilliant and impossible to work with. The letter should make clear whether that's happening.
The closing paragraph gives a direct endorsement. Not hedged. Not "I think they might do well if they work hard." Something unambiguous like "I recommend them without reservation for this program" is what you're aiming for. The reader needs zero ambiguity about where you stand. Length-wise, one and a half pages is usually the sweet spot for academic applications. Two pages if you have substantial material to cover. Anything beyond that is noise. Hiring managers for industry roles often prefer shorter letters, sometimes just a page, because they're scanning dozens of them in a single sitting.
A Problem I Encountered And How I Fixed It
A few years back, a strong candidate asked me for a letter targeting a specific fellowship that required recommenders to address three particular prompts: intellectual curiosity, leadership potential, and resilience under adversity. The fellowship used a structured review process where each criterion was scored separately. I wrote a standard narrative letter first and then realized it didn't map cleanly onto their rubric. The committee would likely miss the key points buried in paragraphs of prose. So I restructured the letter around their three prompts instead of following a conventional flow. Each section became a self-contained argument with its heading, examples, and a clear sentence at the end tying the evidence back to the criterion. It took me about twenty minutes to reorganize, and it probably made a real difference in how the letter was evaluated. Most recommenders don't bother with this kind of alignment, which means a small amount of targeted effort can give a candidate a meaningful edge.
What To Avoid Completely
Never include information that could be verified as false or misleading. I've seen recommenders inflating a candidate's role in a project beyond what's accurate, and it usually comes out during interviews or later references. It damages everyone involved. Also avoid overly personal anecdotes that don't relate to professional or academic performance. Comments about someone's character outside of a work context have diminishing returns and can sometimes raise questions about appropriateness. Don't rely on AI-generated drafts for the full letter. Candidates increasingly submit letters written by language models, and some committees have become adept at spotting the telltale patterns: the overly balanced sentence structure, the absence of messy or specific details, the generic phrasing that sounds correct but says nothing. A brief outline from an AI tool is fine. Turning it into the actual letter requires human judgment and real knowledge of the candidate.
Timing And Logistics That Matter More Than People Think
Ask for the letter at least six weeks before the deadline. If you're asking someone to write about you and you contact them two weeks before applications close, they either write a rushed letter or they decline. A declined request is worse than a late one in most cases because it leaves a gap in the application. Give the person enough time to reflect on concrete examples and draft something that resembles an argument rather than a list of positive adjectives. Provide context materials when you ask. A resume, a summary of the projects you worked on together, the deadlines, and the specific programs you're applying to. The more information you hand over, the less time the writer spends gathering details, and the better the final letter will be. It's a small courtesy that most people skip and then wonder why the letter feels shallow. If you're the one writing the letter, use a simple format that matches what the receiving institution expects. Most universities accept uploaded PDFs or text boxes. Industry roles often use standard letterhead templates. There's rarely a benefit to creative formatting. Plain text on letterhead with your contact information at the top is what professionals expect to see.
When A Letter Shouldn't Be Written
Not every request deserves a positive letter. If you genuinely don't feel confident endorsing the person, the honest move is to say no or offer to write a neutral letter of referral instead. A lukewarm recommendation that reads like a reluctant endorsement is often more damaging to a candidate than no letter at all. Committees read between the lines, and qualifiers like "adequate," "acceptable," or "I suppose they'd be fine" are red flags dressed up as politeness. There are also situations where the recommender lacks sufficient direct exposure. A professor who taught a student in a large course and hasn't interacted with them since is better off declining than writing a letter based on grades and attendance records. The credibility of the entire recommendation system depends on recommenders only endorsing people they've actually supervised or worked closely with.
