I’ve Decided to Apply! What Now?

Once you’ve decided to apply to graduate school, the next step is to identify which programs to target and prepare your application materials.

In this section, we’ll cover key parts of the process:

1. Identifying Schools: how to choose which graduate programs to apply to, including strategies for balancing “reach,” “target,” and “safety” schools.
2. Letters of Reccomendation: how do you approach professors for letters, when is the best time to ask, and who should you ask.
3. Writing Application Essays: how to approach statements of purpose, personal statements, and other written materials effectively.

Each subsection includes practical advice, quotes from students and faculty, and examples to guide your preparation.
More examples are available on the Example Materials tab.

---

Identifying Schools

If you are applying, getting into one or more graduate programs is “infinitely” better than getting into zero. To mitigate the risk, we recommend applying to many programs and applying to a broad range of programs (i.e., programs of varying selectivity).

To see why, suppose, for simplicity, that every school independently has a 10% chance of accepting your application. (See historical Michigan numbers above for comparison; the advice here is primarily drawn from US schools.) If you apply to “only” five programs, there is a (100% - 10%)^5 = 59% chance that the biased coin comes up “tails” each time and you are rejected by all programs. So we want to apply to many programs.

However, even applying to 15 programs with 10% acceptance rates still leaves a 20% chance of being rejected by all of them, so we also recommend applying to programs with a range of selectivities. This usually takes the form of ~10 schools in a band, 1–2 aspirational schools (long-shot selective schools that would be great, but may not accept you) and 3–4 safety schools (less-selective schools that are more likely to accept you).

Students we work with typically apply to 12–16 programs. If you apply to more than 16 programs, you run the risk of failing to secure letter writers or offending your letter writers. While letters of recommendation are primarily shared between institutions, letter writers often specialize them to particular schools, and so it is an increasing imposition to ask them to write too many letters.

In general, graduate programs have become increasingly selective in recent years: few applicants are admitted. For example, the Michigan historical data (see FIXME in this guide) shows master’s degree admission dropping from 15% to 5% from 2014 to 2024 and PhD admission rates dropping from 19% to 10% in that same period. In the opinion of multiple authors of this guide, graduate admissions involves a significant amount of chance. Although our feelings of imposter syndrome may try to convince us otherwise, many strong applicants are rejected and some less-prepared applicants are admitted.

PhD applicants who have been involved in an independent study or undergraduate research project can sit down with their advisors, bring their unofficial transcripts and CVs, and ask for an assessment of where (i.e., what rank) they would be accepted at “on average.” Master’s applicants can do the same with any professor they took a rigorous elective with, or worked as a teaching assistant for, etc., but they may want to schedule the appointment early to make sure that it happens.

Graduate programs are often assessed using nebulous metrics known as rankings. There is much more to a graduate program, and much more to graduate applications, than simple rankings. However, rankings do correlate with selectivity and they can be a useful tool for generating initial ideas about where to apply. The two primary rankings for computer science are:

• US News and World Report: Informally, this ranking of all departments is based on surveys sent to officials at all departments. That is, it tracks perception and reputation (like a “popularity contest”). While this might seem arbitrary, it can be quite useful for many students. If your goal is to get a master’s degree and then enter the workforce, the broad public perception of how prestigious the school is will likely matter more to your future job applications than will the number of research papers published at that school.

• CSRankings: Informally, this ranking is based entirely on bibliometrics: numbers and statistics about how many peer-reviewed scientific publications each department produces in various areas.

One powerful trick is to use CSRankings to identify a school that is ranked higher in the area you care about but lower overall. For example, suppose you are interested in Programming Languages and Software Engineering. In a particular semester, North Carolina State University may be ranked 34th by CSRankings and 51st by US News when all disciplines are included. However, if you turn off all of the areas and then turn on just Software Engineering and Programming Languages, NCSU is now ranked 13th by CSRankings! This suggests that NCSU is particularly strong in PL and SE, but may be weaker in other areas. If you are worried about getting into graduate school, that can be very helpful information. All other things being equal (this is a big simplification), the selectivity of a school is more related to its overall ranking, but the benefit you get out of it may be based on its ranking in your area. So if you know that you want to focus on PL or SE, NCSU may be easier to get into and may offer you a very high quality education once you arrive. (You can repeat this sort of search for whichever areas you personally care about.)

You may find Philip Guo’s vlog on choosing CS graduate schools useful as an additional perspective.

Finally, we note two common pitfalls. The first is restricting attention to a particular location, such as “I will only apply to schools on the East Coast”. You can do a geographically-targeted graduate school application, but we encourage you to double-check this guide’s suggestions on mitigating risks and contingency planning. Second, some lower-ranked PhD programs are easier to get into but are not fully funded. That is a complicated topic of discussion, but the authors of this guide strongly recommend against such programs for PhD applicants. It is typically better to apply again next year than to accept such an offer; PhD programs that ask you to pay your own way rarely have your best interests at heart.

Mitigating and Dealing with Risk

Graduate school applications are inherently uncertain. If you are applying to programs with an average 10% admissions rate, and you apply to 15 programs, “all other things being equal” you have an 80% chance of getting into at least one. But that’s still a 20% chance of receiving no offers of admission.

Planning for potential rejection. If you are worried about ending up with no offers of admission, we recommend discussing things with your advisor. In some cases, it may simply be a tougher year. Graduate school is counter-cyclic with respect to the economy (but at a slight time lag): it could be that overall applications are up this year, and since program space remains about the same, that makes programs more selective overall. It could be that federal funding was down the previous year, and thus programs made fewer offers of admission to PhD applicants. It could also be that you are thinking of applying to schools that are too selective with respect to the average perception of your application strength. In cases like these, multiple contingency plans are possible. For example, you might apply to a wider spread of schools this year, but also be prepared to work for a year and re-apply to graduate schools next year.

It could also be that there is something special or risky about your application. For example, perhaps your undergraduate degree was in one topic (like Psychology) and now you are applying to graduate school in another topic (like Computing). Or perhaps you are restricting attention to schools in a very tight geographical area as one way to deal with a dual career situation. In these situations, the risk mitigation strategy is often specific. A student with a non-CS major might want to establish a record of CS research with a CS publication before applying. A student with a tight locational constraint might want to plan to apply over 2-3 years, rather than just 1.

In general, we suggest applying to a range of schools (including some aspirational schools and some safety choices, in a distribution), applying to at least a dozen schools, applying to industrial positions at the same time (but knowing how to interpret their borderline deceptive short-fuse terminology; see elsewhere in this guide), and making a backup plan for what happens if you do not receive an offer of admission.

Letters of Recommendation

You typically need three letters of recommendation to apply to a graduate program. In general, one of these should be a strong (“primary”) letter of recommendation. The other two can be weaker (“secondary”) letters of recommendation.

The degree to which the letter writers need to “actually” know you varies between Master’s and PhD programs. For a Master’s program, a primary letter from a professor who taught you in a (or multiple) rigorous, upper-level elective that you did a very good job in can suffice. If you have an academic advisor who is one of your professors, this is also an excellent choice. (Recall that Master’s degree programs focus on success in graduate-level coursework.) For a PhD program, your primary letter writer should be someone you worked on a non-trivial project with. The strongest such letters describe undergraduate research projects. However, independent study projects can suffice, as can professors from classes where you did a particularly ambitious job on the final project.

As a general rule, at most one letter-writer can be from industry (e.g., from an internship). Whether or not someone is from industry “shouldn’t” matter, but in practice it has two potential impacts. The first is that some committee members are biased against letters from industry. The second is that some industry letters include the wrong sort of information: what you would include in a reference for someone transferring from one branch of IBM to another isn’t exactly the same as what you’d say to help someone get into graduate school. You can mitigate both of these risks with some early planning. First, we recommend choosing someone from the company that has a graduate degree: people who have gone through the graduate school process successfully are more believable when they say you can do it! Second, we recommend reminding your letter writers of what we are looking for in terms of evidence and argument (see the information on essays and statements of purpose in this guide). If you have only worked in industry a few years, it’s entirely reasonable to reach out to your professors to request a letter. Especially for graduates from smaller programs, your professors haven’t forgotten you or your achievements. In larger schools, professors commonly receive requests from students who have graduated: if you are worried that the professor might not remember you, including a few details (e.g., “I did my final project on XYZ in your class and talked with you during office hours about ABC”) usually suffices.

Any of the following can make a good letter writer. This list is not exclusive.

1. A professor who supervised an undergraduate research project leading to a submitted, peer-reviewed publication. This is the gold standard, and it is rare in general. It is common among applicants to top-ten PhD programs.
2. A professor who supervised an undergraduate research project that you presented as a poster at a professional (or regional undergraduate) conference.
3. A professor who supervised an undergraduate research project that was never actually submitted for peer review.
4. A professor who supervised an independent study project.
5. A professor who taught a class for which you were a teaching assistant.
6. A professor who taught a class for which you were a grader (especially if you stood out or did a good job).
7. A professor who served as your academic advisor and whom you interacted with for more than just one or two advising meetings a year.
8. A professor who was the advisor for an official club (e.g., solar-powered car this, clean water project that, etc.) that you participated in (especially if you stood out or were an officer).
9. A professor who was the advisor for a service activity you were involved in (e.g., broadening participation in computing, helping to get people from all backgrounds involved, etc.), especially if you stood out or were an officer.
10. A professor who taught an upper-level CS technical elective that you did a good job in (especially if you stood out on some element, such as a theoretical exam or implementation project, or if the course is known to be particularly rigorous at your school).
11. A professor who taught a graduate-level CS course that you took (yes, you can sign up for graduate courses as an undergraduate).

Overall, the key factor isn’t the nature of the relationship (e.g., professor in class vs. advisor in club or whatever) but whether or not the person can describe specific situations in which you did things that correlate with success in graduate school.

Many students find that they know who the primary letter writer would be, but they aren’t as certain about the others. You can talk to your primary letter writer about the right strategy for choosing the other letter writers.

Once you have identified potential letter writers, you need to approach them to see if they are willing to do so.

When applying online, you should always waive your right to see your letters of recommendation. When submitting a recommendation, the letter writer can see whether you did so or not. If you do not waive your right, many letter writers (including Wes Weimer, etc.) will not feel comfortable submitting a letter for you. That is because many people view letters as conditioned on confidentiality (i.e., the recommender is providing a reference conditioned on the fact that the contents of that reference will be kept confidential). Some applicants think that seeing your letters is necessary to keep track of who has submitted and by what date. In practice, those functions are unrelated and independent on recommendation systems. You should waive your right to see the content; the system can still tell you if the letter was submitted or not.

Approaching Letter Writers

Many students report that approaching people about writing letters of recommendation is nerve-wracking. We have three concrete pieces of advice for doing so.

1. Approach professors early. Many professors, especially popular professors from larger classes, receive many requests. Professors can typically only agree to a limited number of them while still doing a good job for those who asked first and received an early commitment. (Trying to write 50 letters at the last minute and doing a bad job on each one does not help anyone.) As a result, many professors (including authors of this guide) have an explicit limit and turn down every student once that limit is reached.
   1. While the exact date will depend on your institution, if applications are due on December 15th, asking on December 14th is doomed to failure. For example, Wes recommends to his own students that they approach professors at least two months in advance (e.g., in October).
2. Use phrasing that “leaves a diplomatic way out”. Some people feel uncomfortable refusing students, especially across cultural or language barriers. A person who does not want to write you a letter, does not know you well enough to do so, or does not have time to do so completely, is worse than not having a letter from that person at all. You do not want your letter writer to feel “trapped” or “obligated” into writing a letter: those letters have very stilted, shallow content that is assessed very poorly by admissions committees (who are on the lookout for them).
   1. Wes personally recommends phrasing like “Do you have enough information and time to feel comfortable writing me a positive letter of recommendation?” Note the emphasis on bits like “Would you feel comfortable?” rather than “Can you?”. It’s then easier for the professor to turn you down easily without anyone “losing face”: “I don’t have quite enough information from when you took the class” could mean “I don’t have quite enough information” or it could mean “I do have enough information, but I actually don’t think you’re very good!” In either case, you don’t want that letter, and diplomatic phrasing makes it easier to receive the hint. (Ideally humans would just be direct about this, but not everyone is.)
   2. Kevin recommends similar phrasing, such as “If you can’t commit to writing more letters at this time or don’t feel comfortable writing a positive letter of recommendation, I certainly understand.” Beyond giving your letter writer a graceful “out” if it’s needed, this also demonstrates your respect for the writer’s time.
3. Clearly indicate “Primary” or “Secondary” and “Master’s” or “PhD”. Success in master’s programs and PhD programs is seen as based on different things (e.g., rigorous coursework vs. research). If you don’t specify, the professor may well assume you’re applying to PhD programs (e.g., the professors all have PhDs and may see it as the default). If you are applying for a master’s program and did well in a professor’s class but did not work on research with the professor, the professor may turn you down, assuming that you want a research-describing letter that professor cannot provide. Similarly, many letter writers may turn down students for “primary” letters that they would accept writing “secondary” letters for. A secondary letter does not require as much information or as deep of a relationship. You do not want the professor to think you mean “primary” when you actually mean “secondary”.
   1. Wes personally recommends including this information right next to the “would you feel comfortable” sentence. Example: “Do you have enough information and time to feel comfortable writing me a positive secondary letter of recommendation for a master’s degree program? Professor XYZ will be writing a primary letter for me.”
4. Provide clear, organized, and detailed information for your letter writers. After agreeing to write a recommendation, your letter writers will be doing many things other than writing your letter. Some professors request supporting materials (e.g., a resume/CV, draft statement of purpose, transcript, etc.) as part of your request while others ask for this information after the fact. Having these materials ready to share is both courteous and helps your letter writers craft strong narratives (i.e., emphasize key evidence of the low risk, high reward nature of your application).
   1. Kevin personally recommends sharing a spreadsheet of schools with each letter writer. Only include schools for which you are requesting a letter from the writer (some students will mix and match letter writers), and indicate the degree, program, and application due date in this list. Recommendation requests often get swept up by spam filters, so having a list for cross-checking is very helpful. It also helps for tailoring letters to specific programs.

While all of the application materials that you write or prepare (e.g., your application essays, your transcript, etc.) are due by the application due date, letters of recommendation are sometimes turned in later. Admissions committees understand that you cannot force your letter writers to turn things in faster and do not penalize you for their actions. Website and admission form verbiage will often make it seem like your application will be summarily rejected if any component of the application is late; in reality, your application may be rejected if your materials are late but will not be rejected if your letter writers take another few weeks. See elsewhere in this guide for details and a concrete example. This is difficult for applicants to believe, but it is true.

---

Writing Application Essays

Your application essays are a critical part of your graduate application. The primary essay is often also called your statement of purpose. Elsewhere in this guide, we describe the application evaluation process, including who will be reading your essays and why.

Your application essays are one of the last things you have control over as you apply. You can’t “go back in time” and raise your GPA or finish a research project, but you can polish your essays. Unfortunately, this leads to many stressed students spending time on last-minute changes that are not very impactful.

Based on decades of combined experience across multiple schools (Berkeley, MIT, Virginia, Michigan, UMass, etc.), we identify critical aspects to keep in mind for essays. These are based on the principles of student opportunity cost (you have limited space and each thing you include is something else you cannot include) and that faculty members are making decisions while reading many of these and balancing risk and reward. In summary:

• Length Constraints and Differentiating Information. Many applicants overlook these opportunity costs and spend words on information that is not specific to them. Instead, you should efficiently present information that differentiates you from others.
• Misleading Prompts and Retention. Many applicants misinterpret the webpage prompts for application essays and include irrelevant information. Instead, you should include information that will help your application, including arguing that you are likely to complete the graduate program.
• Persuasive Essays. Many applicants imagine the essays should be descriptive (“I did X and then I did Y”). In reality, the most effective essays are persuasive and convince the reader that you would be a high-benefit, low-cost graduate student.
• Argument and Evidence. Many applicants include unsupported assertions (“I can conduct rigorous analyses”). Instead, you should present concrete evidence leading to such conclusions.

We consider each issue in turn. These apply to both “personal statements” and “research statements”. In addition, this reasoning also applies to what you want in the letters of recommendation written about you.

Applications: Length Constraints and Differentiating Information

Application essays typically have stated length limits. Even if they do not have formal limits, professors may stop reading after 2-3 pages. As a result, you want to spend all of your precious essay words efficiently. It’s not “bad” to say that you like computer science in your essay, it’s just not “as good” (as efficient) as some other things you could be doing. Formally, this is known as opportunity cost. And if two similar people apply to CS graduate school, one of whom spends a sentence on liking CS while the other spends that sentence talking about individual research experience (or whatever), all other things being equal, the admissions committee has more information about the second applicant and will extend the offer to the second applicant. Recall that the people deciding are usually looking for low-risk, high-reward applicants.

We also strongly encourage you to focus on differentiating information (i.e., things that are true about you but that are not true about most other applicants) in your essays. It is relatively inefficient to spend space indicating that you like computer science, that you completed the undergraduate CS major, that the school you are applying to has a good graduate program with many interesting faculty members, or that you enjoyed your upper-level CS electives and want to do more of that. Indeed, speaking personally, Wes Weimer would pay more attention to an essay that said “I am not a CS major, but I am applying anyway, and I want this degree and I think I will succeed: here is why”.

If you were a teaching assistant multiple times and most other students weren’t, talk about that (and what it says about you and why you are likely to succeed at, and complete, the graduate program). If you were a leader in an organization like NSBE and most other students weren’t, talk about that (and what it says about you and why you are likely to succeed at, and complete, the graduate program). If you are a first-generation student, if your Security Project got the highest score in the class, if you were helped by a CS accessibility tool and way to “pay it forward”, if you’ve read a CS research paper (e.g., in an independent study or graduate class) and have ideas about how it could be improved, talk about that. If you did an independent project with a professor, and most other students did not, talk about that. If your internship had you solve an interesting problem for which the right answer was not known, and then had you present the results, talk about that. If you’ve found that you enjoy aspect X of CS more than aspect Y, talk about that with a focus on the positive (e.g., indicate that you have experience with both X and Y, but that you’re particularly passionate about X). (This list is not exhaustive.) Write about the things that are specific to you, rather than common to most applicants, and then indicate what that experience says about how likely you are to successfully complete the graduate program.

In most cases, especially if you are not certain about how to write your application essays, we recommend starting with a straightforward “five paragraph essay” structure (it need not be literally five paragraphs). Your essays should state and support the claim that you believe your experience with X, Y and Z position you to succeed in a rigorous graduate program — a strategy we explain in the “Persuasive Essays” section. Your essay might either lead with the most differentiating information first or lead with the information that is most relevant to your success in a graduate program.

Applications: Misleading Prompts

Consider the following 2025 application prompts from Michigan CSE:

Statement of Purpose: Your statement of purpose should be a concise, well-written document about your academic and/or research background, as well as your career goals. Additionally, it should communicate how CSE’s graduate program will help you meet your career and educational objectives.

Personal Statement: Your personal statement should describe how your life experiences – cultural, geographical, financial, educational, and/or other opportunities/challenges – motivated your decision to pursue a graduate degree in CSE.

Elsewhere in this guide we discuss differences between the two essays. But for now, we note that when professors are commiserating about reading application essays, a common refrain is that far too many essays start with some variant on “Ever since I was a small child, I have wanted to pursue computer science”. This is seen by applicants as responding to “are you motivated to pursue a graduate degree in CS?” In practice, it ends up not being very useful to professors who are evaluating your application. (To demonstrate how professors often interpret these, we elaborate on this specific example. Since most applicants grow up with computers, doing so does not set an applicant apart. Consequently, it doesn’t demonstrate perseverance and commitment like it might have decades ago. Students may think that including this fact demonstrates lower risk or higher commitment because they have “stuck with” CS for many years, but professors may not find it compelling.)

To see why, we reason from application evaluation considerations (detailed elsewhere in this guide). Informally, the “worst” outcome is a student who enters the program but then leaves just before receiving a degree. Applicants often imagine that this could happen because graduate students fail the rigorous coursework or do not pass the preliminary qualifying examination. In practice, those things almost never happen. Instead, graduate students choose to leave because they are unhappy (or, equivalently, because the perceived benefits of graduate school are no longer seen to outweigh the perceived costs). Because graduate programs are so selective (e.g., see historical Michigan numbers), a school that is only accepting 5-10% of applicants can ensure that everyone they accept is likely to succeed. This is often difficult for applicants, in the grips of imposter syndrome, to believe. In practice, almost no one “fails out”. For example, at Michigan CSE, the graduate committee chair indicated that no PhD student had been “forced” to leave after failing the preliminary qualifying exam in the last three years! Formally, this is known as retention (i.e., the retention rate is the fraction of students who accepted the offer and stuck it out through the program to earn the degree).

Instead, students who are not happy in graduate school (see elsewhere in this guide for challenges, such as work-life balance, stress, etc.) choose to leave for lucrative careers in industry.

The chain of reasoning might go like this:

• Faculty members want everyone who is admitted to complete the graduate program and earn a degree.
• Faculty members know the primary reason for people leaving the graduate program is a lack of motivation to continue (unhappiness).
• Faculty members thus want to know, in advance, if a student is motivated or not to complete the program. If there are two applicants and you only have space for one, and one of them is strongly motivated, you are likely to choose the motivated one.
• Faculty members know that one of the primary reasons to get a PhD or a master’s degree is because you want a particular career (such as tenure-track professor).
• Faculty members believe that you are more likely to stay in the program and complete the graduate degree if you think our graduate program will help you attain those career goals. Is this degree a necessary step along your career path?
• Faculty members thus direct the webpage essay prompts to be written with an emphasis on motivation.

Unfortunately, between the start and the end of that chain, the words changed. The prompt should perhaps say something like “Convince us that you will complete the program if you are admitted.” Instead, it says something like “Tell us about your motivation and why our program will help you attain your goals.” This is what is known, in research, as a construct validity error. Every step along that chain of reasoning makes sense. But somehow we end up with prompts that students tend to respond to by writing things like:

I have always loved computer science. When I was a small child, my parents would ask me CS questions. Your highly-ranked program is so strong and well-respected that it will definitely help me attain any CS career goal I might consider!

When writing, we strongly encourage you to correctly interpret prompt text about motivation, program strength and career goals as “Convince us that you will complete the degree program (because it aligns with your career goals)”. The prompt says motivation, but you may be better served by addressing completion or retention. Do not spend your precious, limited pages telling the admissions committee that their graduate program is strong: they already know that. Do not spend your precious, limited pages waxing poetic to a CS admissions committee about how much you like CS (they can guess: you’re applying to CS grad school).

Applications: Persuasive Essays

Many applicants assume that the essays they write should be descriptive. This is understandable: the word “describe” is often present in the prompt (“Your personal statement should describe how your life experiences”). However (as discussed elsewhere in this guide), that can be misleading.

Your application strongly, strongly benefits from being persuasive. That is, your application essays should make the explicit argument that you are a high-reward, low-risk applicant who is likely to succeed at, and complete, the graduate program.

The most common argument structure is, implicitly, that the past predicts the future. Broadly, graduate students carry out certain activities (depending on PhD vs. Master’s track, etc.):

• Taking graduate classes
• Conducting independent research
• Assisting with teaching
• Serving on local or international committees

The most direct argument is that you have prior experience and success with some subset of those activities and are thus likely to succeed at them in the future. For example, if you were a very effective teaching assistant as an undergraduate, you will probably be a very effective teaching assistant as a graduate student. If you submitted a paper to a peer-reviewed venue as an undergraduate, you will probably be able to do so again as a PhD student. And so on.

Note that this is not an if-and-only-if. You could have zero research submissions as an undergraduate and learn how to write successful research papers in graduate school (that was true for Wes Weimer, for example). However, an applicant who has already demonstrated success at a graduate school task is usually seen as lower risk. (See below for how demonstrating a skill or experience that is relevant to a graduate school task can be just as compelling: you do not need to have done research per se to have a competitive application.) If a new faculty member only has money to admit one PhD student, or if a master’s degree program only has one more open slot, ties may be broken in favor of the lower-risk, higher-reward applicant.

You are thus more likely to be selected if you convince the evaluators that you are a lower-risk, higher-reward candidate.

While simple on the surface, structuring your application materials as persuasive (rather than descriptive) is arguably the single most important piece of advice in this guide. Consider a professor reading dozens of application essays in sequence. The eyes start to glaze over. Attention is hard to maintain. Many essays say roughly:

I was a teaching assistant for Discrete Math. I got a perfect score on my OS project. I did my senior thesis on a literature review of ethical AI.

Such essays tend to blend together: an essay that “merely” lists qualifications blends in among many other qualified applicants. By contrast, consider reading one that makes a persuasive case:

As part of my undergraduate studies / years as a junior developer, I have gained experience generating creative ideas, carrying them out rigorously, and communicating those results to others. I believe my experience with an AI project, OS implementation, and as a teaching assistant position me well to succeed in YOUR_SCHOOL’s rigorous graduate program. I demonstrated creativity while working on my senior thesis AI project by doing ABC. In an OS project, I practiced skills DEF and ultimately passed all of the held out tests, demonstrating my ability to focus on correctness. Finally, as a discrete math TA, I was responsible for taking abstract ideas GHI and conveying them to a broad audience. I believe this preliminary experience with creativity, rigor and formal communication will help me succeed in graduate school.

This mimics the persuasive, logical style of many research papers. To new students, spelling out all of the links can seem soulless or rote. Unfortunately, a tired faculty member reading many of these essays may not make the jump between “This essay is telling me about someone being a teaching assistant” and “This applicant is good at communication, and the evidence for that is …” In high school or undergraduate composition classes, we are often encouraged to avoid simple structures like these. But starting with them can help us with organization. And there is still room for style: see the example application materials for examples of essays with personality and hooks and heart. We strongly recommend that you start with the structured argument and add heart to it, rather than starting with heart and hoping the reader will be persuaded.

Applications: Argument Structure

In application essays (and other documents, such as CVs or letters of recommendation), arguments that drill “one level down” to provide an anecdote or evidence in support of a claim are viewed as more persuasive.

In broad strokes, this may seem simple. “I am good at math” is slightly less persuasive than “I am good at math, and the evidence for that is that I earned an A+ in my calculus class.” In practice, there are two aspects of this that may merit consideration as you write your essays and CVs: “how well” and “specific activities”.

First, skeptical readers will often assume that you are not directly lying in your essays or CVs (that would often be easy to cross-reference with your reference letters, etc.) but that you are already putting your best foot forward. For example, if you earned an “A” in operating systems and a “B” in graphics, you might only spend sentences in your essay talking about the “A” in operating systems and what it says about your likely future success. You are not lying: the “B” in graphics is available on your transcript, etc. But your role is to be an honest advocate for yourself in your essays and CVs, not to cut yourself down. Admissions committees will already be looking for reasons to reject you (or to sort someone above you who seems less risky: see the Paradox of Yield in this guide).

An implication of this is that if you write something circumspect, perhaps out of modesty, many readers will assume you have written the strongest claim possible and judge you based on that.

For example, if your essay or CV indicates that you were a teaching assistant for data structures for one semester, and that is all you say, a skeptical reader (or someone looking for the lowest-risk candidate to fill one open slot) may assume that you did a bad job as a TA and that you were never invited back. After all, the reasoning goes, if you did a good job, why didn’t you say that? And if two people apply, one of whom says “I was a TA once” and one of whom says “I was a TA once, and here is some evidence that I did a good job”, the second candidate is lower risk. Sure, the first candidate might also have been a good TA. But we weren’t given any evidence of that (informally: “and I’m reading dozens of these applications at a time, and I’m tired, and it’s not my job to make your arguments for you, etc.”).

In our experience, students are very very good at describing what they did. “I was a summer intern at Microsoft in the Visual Studio group.” “I was the president of our solar-powered car club.” “I was a teaching assistant for discrete math.” We strongly encourage you to intentionally add a second sentence to each essay line or CV line that follows up with how well you did.

The good news is that you have a huge amount of creative freedom in what argument to make. Some success metrics are common. Did the group at Microsoft make you a conditional return offer, or did the discrete math professor ask you to come back and be a TA again? Not everyone earns a return invite. But there are other options. For example, perhaps your solar car club won a regional competition. Or perhaps club attendance was higher under your leadership than in prior years. Or perhaps you secured a faculty sponsor for your club, unlike prior years. Or perhaps three more senior people in the club told you they used club experiences in their job interviews and got jobs. You could position any of that as evidence of success. But the evaluators can’t guess it: you have to say it explicitly.

Application Essays: Evidence To Include

There are many specific activities relevant for graduate school success. You have creative freedom about what you pick. Elsewhere in this guide, we described high-level categories like taking graduate classes, doing research, and helping with teaching. Applicants often feel like they haven’t really done any of that, so how can they write compelling essays? Well, you may not have written a scientific paper per se, but you may have carried out one of the specific sub-activities of writing a scientific paper in another context. And your success there can be used as evidence that you are slightly more likely to succeed at graduate school. For example, graduate students may do any or all of the following (this list is not exhaustive):

• Succeed at rigorous, mathematical courses
• Succeed at qualitative paper-reading courses with analysis and presentation components
• Succeed at large project courses that involve writing code in teams
• Identify impactful research problems in the real world
• Combine two prior techniques to make a new idea
• Generate falsifiable hypotheses and design controlled experiments to carry them out
• Conduct human studies with Ethics Board (IRB) approval
• Write formal (or machine-checkable) proofs of theoretical claims
• Perform statistical analyses of large data sets
• Apply machine learning or AI models to understand patterns in real-world data
• Generate effective figures, tables and visualizations to convey formal results
• Write up results for a scientific or broad audience
• Give presentations on research results and answer live questions about research techniques
• Lead discussion sections to assist students
• Conduct office hours to assist students
• Revamp or update assignments to improve accessibility for all students
• Mentor junior students in technical skills

In broad strokes, graduate students might benefit from any of the following properties (this list is not exhaustive):

• Creativity
• Patience
• Rigor
• Diligence
• Humility
• Communication
• Leadership
• Responsible
• Analytical

Something that may not be evident to students preparing to write essays is that you can often reorganize the evidence and argument structure any way you like, based on what you have done. For example, your evidence demonstrating creativity might be that you came up with a research topic for your independent study project. However, your evidence demonstrating creativity might instead be that when your solar car club was facing a room shortage on campus, you came up with the idea of approaching local companies for storage space. Similarly, your evidence demonstrating diligence might be your manual annotation of a large dataset as part of a qualitative analysis for a research project. However, your evidence demonstrating diligence might instead be your literature search for a potential project and how you used three different techniques to find overlooked prior work.

The planning stage for an application essay might begin by listing your experiences and successes and learning opportunities. “I was in club X”, “I did well in course Y”, “I helped with project Z”. Then, in a separate column, you might list some properties of a successful graduate student. You’ll eventually want about three, but you might start with for. Then, you make links (sometimes physically drawing this on the whiteboard or on a tablet can help) between the activities (evidence) and the properties you want to persuade people you have. Usually this is a many-to-many mapping. You could use your TAing experience for discrete math to demonstrate your effective communication, but it might also be used to argue that you can understand formal topics rigorously. Ultimately, you pick the three aspects of yourself you are trying to get across and match up your available evidence to them in a somewhat balanced manner. If it turns out that one of the properties has no evidence, reconsider the high-level structure (e.g., pick different aspects of yourself to champion).

To restate aspects of this one more way, if you are not certain how to convince someone that you are Creative (for example), inline the definition. If I am writing a letter of recommendation and I want to argue that Claire is creative, it is not very compelling if I just assert it (“Claire is one of the most creative students I have seen!”). But if I talk about what creativity means (what the steps or specific activities are), it is seen as more compelling. For example:

I once had the opportunity to observe Claire’s creativity in a research setting. We were faced with a new problem with no established solution: how to assess automated program repair algorithms. First, Claire generated four candidate solutions with various tradeoffs, from human trials to studies of archival data. Next, she designed a “fail early” experiment for each one: a way to try it out quickly to see if it would likely be fruitful. She tried out those evaluations on her own and reported the results back to me. Claire concluded that three of the approaches were very risky and unlikely to work, and proposed that we pursue the fourth. I found her argument compelling, and she used that solution in the paper, which was eventually accepted for publication.

That example is a bit “on the nose”, but it does show how “X was creative” can be defined in terms of specific activities (generate candidate solutions, evaluate candidate solutions, choose a solution to go forward), some or all of which can be associated with the applicant. You can use the same approach when making the claim, in an essay, that you have demonstrated creativity. This can also be a good way to describe a non-traditional approach or an idea from a different discipline. Projects, research or activities associated with other fields can be very strong evidence, since you are assessed more on the path and actions you took than on the exact area of the final outcome. You don’t need direct research to claim that you are creative (or rigorous, or …): instead, inline the definition of creativity (or rigor …) into components and present evidence that show those components.

This approach works for other rich, nuanced and desirable properties as well. For example, you could just argue that you are an effective instructor based on your TA experience. But if you break that down into and describe a time when you make an extra practice worksheet for the students; a time when you presented a complicated topic with code, prose and a diagram; your habit of giving students space to ask questions rather than you moving on immediately to the next explanation; and ultimately how your discussion section and office hours saw the most students as a result – that inlined definition is more compelling. It gives the reader more confidence that you will do at least part of that again in the future, and thus that you are a lower-risk and higher-reward applicant. (In all of this, you get to pick what evidence and structure you use! If you never did a code-prose-diagram thing to explain something, but instead you made a little game with flash cards, use that instead! Strong essays are often constructed by enumerating all of your evidence, enumerating the claims you would like to make, lining them up, and then writing the whole thing with a clear, persuasive structure.)

Dividing Material Between Two Essays

You often have a degree of freedom about which aspects of your experience to put in the research-focused essay and which to put in the personal-focused essay. In practice, this does not matter as much as applicants may think it does. Evaluators will typically read them back to back. Many experiences or skills could fit in either one. If you helped organize a tutoring club and made practice problems for other students, that could fit well in a more personal statement. But it could just as well be placed in a research-focused essay, where it serves as evidence that you can rigorously synthesize and analyze ideas (turn course concepts into practice problems) and communicate them to a broad audience (convey them to people who are not already experts) — both of which can help with graduate-level research or coursework. We often encourage students to list out all of the experiences, properties, skills, etc., that they wish to convey, and then divide them up between the two essays. The exact placement of which point goes in which essay is not as important as having essays that persuasively argue that you are a high-benefit, low-risk candidate who is likely to complete a rigorous graduate program and use that training to achieve your impactful career goals.

Putting it all Together

Application essays are persuasive documents that use evidence and argument to position you as a high-reward, low-risk candidate who is likely to complete a rigorous graduate program.

You only have so much room: use your essay words to describe yourself specifically. Prompts often use phrasing about motivation: focus on career goals and why you will complete the program. Do not merely state what you did (“I was a teaching assistant”) or claim things about yourself (“I am creative”). Instead, lay out logical arguments that: (1) you possess certain properties (skills, habits, etc.), because (2) you have done certain things (experiences, projects, etc.), and (3) you succeeded at them (positive outcomes). This helps evaluators gain confidence that you are likely to apply those skills and habits in graduate school as well, and are thus likely to succeed there.