A major feature that makes applying for NSF funding unique is their use of external review panels. NSF reviews are conducted primarily by panels of external reviewers who don’t work at the NSF and have other full-time jobs. While these reviewers are representative of your scientific peers, they are a unique audience.

The average reader for a research paper…

• chooses to read the paper
• is an expert in the (sub)field of the paper
• wants to learn something

The average reviewer on an NSF panel…

• agrees to read it on top of their normal workload
• is from a related field, but probably not an expert on your topic
• wants to finish as quickly as possible while being fair

As a result, you should not write an NSF proposal like a research paper or preregistration of a study. First, you won’t have the necessary space because you’re restricted to 15 pages. This restriction is why I describe writing an NSF proposal as a trust-building exercise with the reviewers. You simply cannot put all of the information that they would need to evaluate the project in 15 pages; instead, aim to build trust that you know what you’re doing. A big contributor to building trust is to be specific about your goals, outcomes, and the path between the two. I always aim to be one level more specific than I, as an academic trained to hedge my statements, feel comfortable with. It’s expected that the small details of the project might change over time, but being specific about the details, like materials and assessments you’ll use, who you’ll work with, and what effects you expect, shows that you have thought through the project and how to implement it.

Though specificity helps to build trust, remember that reviewers aren’t intrinsically motivated, nor probably equipped, to slog through 15 pages of dense text full of jargon and subfield-specific arguments. Instead, your proposal should look more like a magazine article than a research paper. The goal is to direct the reviewers to the most important points and provide supporting details. Below are features that make reviewing proposals much easier.

• Include something visually salient on every page. Reviewers probably won’t read your proposal top-to-bottom in one sitting, so make it easy for them to remember what the main point of each page is by using figures, tables, bulleted lists, or even bolded sentences.
• Make information easy to find. Reviewers have about 6-12 proposals to read. During panels, they are often trying to find details from individual reviews to refresh their memory or address a concern another panelist has raised. Make this process easy by using many, meaningful section headings. I’ve also become a big fan of putting methods or assessments in tables or bulleted lists rather than text.
• Include a logic model. A logic model is typically included in the evaluation section because it shows the path from the resources of the project to the activities to the outcomes to the impact. Thus, it is a tool for evaluating the progress of the project, and it can also be a tool for summarizing the goals and methods of the project.

• Include examples. Because reviewers are not likely in your subfield, use examples that demonstrate the potential impact of your project or the problem that you are solving. Otherwise, reviewers might struggle to contextualize your project.
• Include interdisciplinary expertise. The best NSF projects bring together different areas to create something new. While CAREER is a solo-PI program, you can include advisors to provide interdisciplinary expertise. It’s especially compelling in CAREER to include advisors who will help you to develop a new skill that will accelerate your career trajectory.
• Don’t oversell your contribution. Though many junior PIs seem to think that they need to overpromise outcomes to be competitive, this practice actually makes them less competitive. Be clear about the contributions that you intended to make, show that they are feasible in the timeline of your project, and also be clear about what work will be left to do at the end of 5 years. Especially don’t oversell contributions related to DEI topics.
• Include a timeline. A timeline shows which activities you will do in each year of the project. It will show whether you are trying to do too little or too much.

• Don’t mis-cite references. The reviewers are scientists. Even with limited time, at least one reviewer will get curious about something you say and look up the citation for it. Make sure it says what you say it says.
• Highlight connections to solicitation goals/NSF rules. If you are making a decision because of a program goal stated in the solicitation or rule in the PAPPG, say it. The reviewers won’t have memorized either of these documents.

These are all things that can make a reviewer’s life easier. Most of them are also things that I wouldn’t do, or not to the same extent, in a research paper. Recognize that grant writing is a related but separate skill from the type of writing we typically practice through our early career and give yourself plenty of time to write and revise. For more information about submitting to NSF CAREER, check out Structure of an NSF Application or CAREER-specific advice.