Career Advice for Policy Researchers

I’ve been a policy researcher (aka “think tanker”) for eight years.^[1] Over this time, I’ve learned how good scholars conduct and share their research, build and manage their teams, and push their ideas into public policy. I’ve also seen some scholars struggle—falling prey to partisanship, corruption, or simply poor research or management practices.

This post lays out my thoughts on how to be a good policy researcher. The first section reviews useful skills, practices, and behaviors. The second section explores how to write reports, build useful tools, engage with journalists, and host events. The third and last section examines common failure modes by which policy researchers lose productivity and influence.

I hope this post helps to inform new and aspiring policy researchers on how to succeed in this field, and perhaps even remind established scholars of best practices.

TL;DR: 

• Learn to read, write, and communicate like the best policy researchers and advocates, while also developing your discipline, teamwork, and creativity.
• Write papers, build useful tools, speak with journalists, and host productive events designed to push good ideas into public policy.
• Avoid common failure modes for policy researchers, such as forgetting your audience and their particular needs, failing to operate at the speed of relevance, corruption, or writing in a bubble.

I. Personal: What are useful skills, practices, and behaviors?

The best policy researchers…

• are good writers and write a lot.^[2]
• are constantly learning and trying out new research tools and methods.^[3]
• are secure enough in their intellect to ask “dumb” questions.
• make clear, transparent arguments with appropriate levels of nuance.^[4]
• regularly present their research and welcome constructive feedback.
• establish a specific area of expertise, whether regional (e.g., China), functional (e.g., nuclear weapons), or both (e.g., Chinese nuclear weapons).

You get additional points for…

• reading widely and refusing to be siloed in a single field of study.
• avoiding common traps of expertise, e.g., overestimating the average person’s familiarity with their field, or misplaced confidence in forecasting abilities.
• being comfortable operating at times as an “advocate”, i.e., having good networks in the policy community, taking positions in policy debates, and generally making things happen.
• being a kind person who’s easy and fun to work with.
• studying the history of your discipline and best practices, i.e., reading stuff like this.

These qualities are generally understood and appreciated. There are a few other qualities, however, that I think are undervalued but critical for effective policy researchers: discipline, teamwork, and creativity.

Discipline

A disciplined scholar commits adequate time to research projects. Although their attention is often pulled in a hundred directions—on events, external relations, fundraising, management, and various other tasks—they schedule blocks on their calendar for deep work on longer publications for which they discourage any interference. And while they enjoy reading deeply, they know when it's time to start writing—and can shift back-and-forth between modes of consumption and production.

A common problem across think tanks is for teams to accumulate research debts because leaders are swamped with networking, fundraising, and various other non-research duties. Leaders therefore become bottlenecks in their research pipeline by failing to put enough words on paper or provide sufficient feedback to staff who are doing so. Better discipline and time management is the only cure.

Teamwork

Policy researchers are not solo acts. Unlike academics, they are evaluated more on impact than the quantity of first-author publications, and, as my RAND colleague Melanie Zaber explains, “Impact usually happens because of collective effort and insight—so we value teamwork.” Even the best policy researchers must work with a team of researchers and administrators to be effective. They don’t need to be extroverted, but they can’t be loners. Think nerds, not dorks.

Good think tankers therefore value project management and appreciate the importance of regular team communication, tag-ups with project sponsors, tracking budgets, and submitting work on time.

They're also good networkers. They present at conferences and workshops, and chat with policy stakeholders (and potential donors) across government, industry, and foundations. They track which members of Congress work on their portfolio. They maintain a network of advisors who can review and discuss papers before they're published. They might even use this network to assist in fundraising or to write short publications of their own.^[5]

As a freelance editor, I frequently review reports that have already been reviewed and marked up. There’s a huge difference in writing quality between papers that have 1-2 polite reviewers and those that have 3+ reviewers with subject matter expertise who carefully analyze and wordsmith every line.

Creativity

Good think tankers value aesthetics and storytelling. They know that policymakers have little patience or interest in reading unappealing, jargon-filled reports, so they carefully consider report design, layout, and readability. They also recognize the utility of tools, games, and graphics (see section below).

A senior defense researcher produced a unique and high-quality assessment on U.S.-China military competition for a top Pentagon official. The sponsor told the researcher that it was useful, but a bit hard to read, and could use more graphics. The researcher recruited a graphic designer to support a second version of the report and soon after DoD and congressional staff began reading and sharing it widely.

II. Products: What do you write, build, say, and organize?

Reports

A policy researcher’s primary product is an analytic report. It should focus on important, tractable, and neglected (ITN) issues. Most reports should offer actionable recommendations. That means:

• Identify the individual or organizational actors that can implement your recommendations.
• Estimate the financial and personnel requirements.
• Acknowledge political constraints and relevant policy tradeoffs.

Sometimes authors may not write actionable recommendations if their primary objective is to frame debates, set policy agendas, or provide neutral assessments. In my CSIS report on countering small drones, for example, I provided a list of metrics that the U.S. Department of Defense should track to evaluate program success. However, I think researchers tend to overvalue these kinds of analyses when policymakers are looking for specific recommendations or even draft legislative language.

As New America CEO Anne-Marie Slaughter explains, “What we are selling is not just ideas. Its ideas married to action for the purposes of change.” In other words, we should avoid saying interesting things for the sake of saying interesting things.

Consider the following suggestions when writing longer reports: 

• Write out your research objectives. Try to make them SMART: Specific, Measurable, Achievable, Relevant, and Time-bound.
• Include an executive summary. Bullet points are better than abstracts. Organize with the Bottom Line Up Front (BLUF), and selectively bold, italicize, and/or underline as appropriate.
• Make your report skimmable. Your table of contents should link to the respective sections, and your PDF should be OCR’d to enable key word searches and copy/pasting. You should not include multiple dense pages of text without some kind of break via headers, graphics, or pull-quotes.
• Create short derivative products to highlight takeaways and recommendations. Common options include a short commentary, two-pager, or briefing slides. Again, bullet points are often preferred.
• Conclude your report with (or separately develop) a list of research questions that you would like to tackle later if given enough time and money. This can help with future proposals.

Tools, games, & graphics

Think tankers also build tools, games, graphics, and other resources to support the policy community and help educate the general public. Consider the following: 

• Develop educational websites like AEI’s Defense Futures Simulator, which allows users to “create, simulate, and analyze alternative strategies and budgets” for the U.S. Department of Defense.
• Organize online communities like Bellingcat's Discord, which connects hundreds of open-source analysts around the world.
• Host (war)games,  tabletop exercises, and simulations to inform analysis or educate participants.
• Create research distillations, perhaps in the model of Open Philanthropy’s Living Literature Review, and graphics that help translate research findings faster and spread them more widely.^[6]

Media engagement

Media engagement should be a high priority for policy researchers. Even if our research is brilliant and our reports are beautiful, they won’t automatically attract readers upon publication. Rather, our readership often depends on popular media outlets sharing and summarizing our findings to a wider audience. Think tankers and journalists thus enjoy a symbiotic relationship wherein think tankers provide background for stories, and journalists direct readers to policy studies and recommendations.

Conversing with journalists is also healthy for policy wonks. Interviews force you to effectively summarize your work and revisit basic questions: Why does your research matter? What are the main takeaways? How do you know what you know? You should obviously think about these questions throughout your research process, but speaking on-the-record adds a personal cost to failure. It puts skin in the game.

You should, however, also know when not to engage with the media. Respectfully decline offers to speak on subjects for which you are not an expert, or on breaking news that you’re not actively tracking. As P. W. Singer cautions, do not “confuse visibility with utility.”

Events

Many think tankers—especially at the junior levels—are part-time event planners. They might organize large rollout events for new reports, or host smaller roundtables or book talks.

Event planners and moderators should:

• Prepare well in advance. Moderators should draft several pages of questions—far more than is necessary—and rehearse and with panelists to ensure a seamless conversation.^[7]
• Advertise early. Send personal invitations to policymakers and journalists. Email an event reminder ~48 hours in advance.
• Make a detailed Run of Show. It’s a good practice to write down and mentally walk through event day to ensure full team awareness, avoid missing things, and anticipate potential issues and proactively plan to prevent them.
• Expect the unexpected. Be prepared for everything from fire alarms to protesters to electrical outages. Keep support staff on standby and save phone numbers for relevant teams (e.g., security, AV).
• Encourage debate among panelists while keeping conversations cordial and, when necessary, clarify the speakers’ points of agreement and disagreement.
• Avoid opening remarks and be aggressive in fielding audience questions. You can and should politely speed up or reinterpret a rambling question from the audience.
• Get the little things right. People will notice and appreciate things like name tents being aligned perfectly parallel across the table, good stapling, and well-timed coffee and refreshment breaks.
• Avoid organizing large event panels (4+ people) if you want to engage speakers beyond simple talking points. Similarly, avoid hosting large workshops if you’re seeking substantial, actionable feedback.

III. Failure Modes: What should you avoid? 

A failure mode is “a potential way an initiative or project could fall short of its goals, leading to a significant loss of impact.”^[8] 

Michael Aird explains six ways that research projects can go wrong: 

1. It's simply inaccurate.
2. It's unclearly written.
3. Accurate. Clearly written. But irrelevant to any important topics.
4. Accurate. Clearly written. But irrelevant to any important decisions.
5. Accurate. Clearly written. But hard for key decision-makers to use.^[9]
6. Accurate. Clearly written. But never seen by key decision-makers.

I'd add one variant to #6: if a policy choice is time-sensitive, you fail if you don't provide your research to decision-makers before they make their decision. I think this particular failure mode is very common. Think tankers frequently operate too slowly to be relevant to urgent policy decisions. They may be slowed down due to internal bureaucracy (e.g., slow publication timelines), external bureaucracy (e.g., project sponsors are slow to respond), or classification burdens.

In addition to Michael's list, I've found three more broad failure modes for researchers: forgetting their audience, corruption, and writing in a bubble.

Forgetting your audience

You should always remember your audience, no matter what you’re writing, building, or organizing. Your primary audience for a single product should be one or a few (never all!) of the following: 

1. Policymakers — convince a decision-maker to take certain action(s).
2. Journalists — inform journalists so that they can convince their wider audience on the value of a certain policy.
3. Researchers — recruit fellow researchers to commit additional attention and resources to your policy topic.
4. Public — educate large numbers of people whose collective opinion can push decision-makers to take action.

Corruption

A think tanker’s reputation is everything. A reputational hit stains their home institution and hollows their policy influence. Good scholars are therefore extremely attentive to their reputation and potential conflicts of interest. They know how to pitch and lead projects while maintaining research independence, regardless of the funding source. They are transparent about their donors, and their publications acknowledge any direct sponsorship. 

Bear in mind the words of Frank Sacker from Showtime’s Billions, who warns that “Principle doesn't usually go away all at once. It's a creeping erosion.” In other words, an ethical think tanker does not simply wake up one day and decide to accept luxury handbags from a foreign government intelligence officer.

Writing in a bubble

Policy researchers need to get out of the building and connect with stakeholders across government, media, academia, and industry to hear diverse perspectives on their subject. They should involve decision-makers throughout their project timeline to help identify political constraints or interests.

Writing in a bubble also includes failing to ask for and incorporate feedback. Unfortunately, many researchers will follow the norm of requesting feedback, but then object if the feedback if negative or suggests time-consuming changes (e.g., using a different methodology, or significant changes to the report structure). You can avoid this dynamic by seeking feedback throughout your research process instead of towards the end when you've finished a first draft.

Conclusion

I’m confident that new and aspiring think tankers who follow these principles will be successful. To recap the main points: 

• Learn to read, write, and communicate like the best policy researchers and advocates, while also developing your discipline, teamwork, and creativity.
• Write papers, build useful tools, speak with journalists, and host productive events designed to push good ideas into public policy.

• Avoid common failure modes for policy researchers, such as forgetting your audience and their particular needs, failing to operate at the speed of relevance, corruption, or writing in a bubble.

   

This post was written in a personal capacity and does not necessarily reflect the views of my current or past employers/clients. Thanks to Reja Younis, Christian Ruhl, and Farah Rishi for reviewing and providing helpful feedback.

1. ^{^}

   I’m currently a defense analyst at RAND and freelance editor for several policy-focused think tanks and foundations. Previously, I was a deputy director and fellow at the Center for Strategic and International Studies (CSIS).

2. ^{^}

   My favorite writing advice is by George Orwell (1946) and Scott Alexander (2016).

3. ^{^}

   Nowadays, they’re the ones experimenting with how to incorporate LLMs into their research process.

^{^}

See also epistemic legibility.

^{^}

Sometimes these advisors are formally recognized as “non-resident associates.”

^{^}

I think these products are undervalued. Check out Chris Olah and Shan Carter’s excellent article on Research Debt and the importance of distilling what is known.

^{^}

Certainly, too much rehearsal can be bad—you don’t want your events to sound like panelists are reading off a sheet of paper. But at minimum, go over the operations and introductory questions with panelists.

^{^}

Definition via Gemini AI, with minor edits by author.

^{^}

You fail if your product is not responsive to the decision-maker’s format preferences, such as reports, two-pagers, briefing slides, videos, podcasts, interactives, or personal briefings.

Show all footnotes