The Heilmeier Catechism is a set of questions that George H. Heilmeier, Director of the Defense Advanced Research Projects Agency (DARPA) from 1975 to 1977, formulated to pressure-test research proposals before DARPA committed funding to them. DARPA still publishes the catechism on its own site as guidance for anyone pitching a program, and it has since spread well beyond DARPA into other federal research offices, university proposal-writing workshops, and industry R&D review boards. Answering the questions in plain language, before a solicitation forces the issue, is one of the fastest ways to find the weak point in a proposal’s logic.
A note on the count. DARPA’s own published version lists eight questions, not nine — this page verified that directly against DARPA’s official Heilmeier Catechism page. Some secondary treatments (grant-writing workshops, proposal-review guides) present a “nine questions” version by splitting the fifth question — “What are the risks and the payoffs?” — into two separate prompts (risks; payoffs), since it is really asking two different things. That split is a common and defensible paraphrase, not an official DARPA addition, and this guide treats it that way: the canonical eight questions below, with the risk/payoff split called out explicitly so a proposal team can answer it as either one question or two, whichever their solicitation’s page-limit and reviewers expect.
The eight (or nine) questions, in order
| # | DARPA’s question | What a reviewer is actually checking |
|---|---|---|
| 1 | “What are you trying to do? Articulate your objectives using absolutely no jargon.” | Can you state the goal in one or two plain sentences a non-specialist program manager can repeat back accurately? |
| 2 | “How is it done today, and what are the limits of current practice?” | Do you actually know the state of the art, or are you attacking a strawman? |
| 3 | “What is new in your approach and why do you think it will be successful?” | Is there a real technical insight here, or is this the same approach with a bigger budget? |
| 4 | “Who cares? If you are successful, what difference will it make?” | Does the outcome matter to someone beyond the proposing team — a mission owner, a field, an agency? |
| 5a | “What are the risks… | Have you identified the ways this fails, not just the ways it succeeds? |
| 5b | …and the payoffs?” | Is the payoff big enough to justify those risks? |
| 6 | “How much will it cost?” | Is the budget realistic and does it match the scope you just described? |
| 7 | “How long will it take?” | Is the schedule realistic, and does it match the technical difficulty implied above? |
| 8 | “What are the midterm and final ‘exams’ to check for success?” | How will anyone — including you — know objectively whether this worked, before the money runs out? |
Last verified against DARPA’s published Heilmeier Catechism page: 2026-08-16.
Worked example: answering all eight questions for a hypothetical proposal
The example below is an illustrative composite built to show the pattern of a tight, jargon-free answer at each step — it does not describe a real funded program, institution, or investigator.
- What are you trying to do? Build a sensor that tells a warehouse robot which boxes are damaged before it picks them up, without needing a person to label training images by hand.
- How is it done today? Human inspectors flag damaged boxes, or robots rely on cameras trained on thousands of hand-labeled example images that take months to collect and go stale every time the box design changes.
- What’s new in your approach? A method that learns what “damaged” looks like from a small number of examples by comparing a box’s shape against its own shipping manifest, instead of learning from a large labeled image set.
- Who cares? Every warehouse operator currently paying for manual inspection or re-training a vision model every time packaging changes.
- What are the risks and the payoffs? Risk: the shape-comparison method may not generalize past the box types tested. Payoff: if it works, warehouses could deploy the system on new product lines in days instead of months.
- How much will it cost? [Filled in with a real, itemized budget in an actual proposal — omitted here since this is an illustrative composite, not a real costed program.]
- How long will it take? Eighteen months: six to build the method, six to test it on real warehouse lines, six to document failure modes.
- What are the midterm and final exams? Midterm: detect damage on box types not seen during training at better accuracy than a human inspector’s error rate. Final: run for 30 days on a live warehouse line with fewer false negatives than the incumbent manual process.
Why the catechism works as a proposal-writing tool, not just a review checklist
The questions are ordered deliberately. Question 1 forces plain language before anything else is allowed to be technical, which catches proposals that are hiding a weak idea behind jargon. Questions 2 and 3 force a genuine comparison against the current state of the art, which catches proposals that don’t actually know what already exists. Question 4 forces a real answer about impact before the proposal is allowed to talk about budget or schedule, which catches technically sound work nobody actually needs. Only after those four are answered honestly do cost, schedule, and success metrics (questions 5-8) become meaningful — a budget and timeline attached to a vague or unimportant goal is not useful information to a reviewer.
Because of that ordering, teams get the most value from the catechism by drafting answers to all eight questions before starting to write the formal proposal narrative, not by retrofitting them into a draft that already exists. A proposal where the cost and schedule sections were written first, and the “who cares” answer was added afterward to justify them, tends to read that way to an experienced reviewer.
Where the catechism is used beyond DARPA
DARPA program managers use it internally when screening ideas before a formal solicitation is even issued, and it appears in guidance for DARPA Broad Agency Announcement (BAA) white papers, where a tight, jargon-free answer to the first four questions is often what determines whether a program manager invites a full proposal. The same eight-question structure, or a close variant of it, has also been adopted well outside DARPA: other Department of Defense research offices use similar screening questions (see this site’s overview of the DoD research funding landscape across DARPA, ONR, AFOSR, and ARO), and university research-development offices commonly teach it as a general proposal-clarity exercise regardless of which agency a researcher is ultimately targeting — including for agencies with a very different funding model, such as the intelligence community’s IARPA.
Frequently asked questions
Who created the Heilmeier Catechism?
George H. Heilmeier, an engineer known for early work on liquid crystal displays, formulated the questions while serving as DARPA Director from 1975 to 1977. DARPA continues to publish them as guidance for anyone proposing a program to the agency.
Does the Heilmeier Catechism have eight questions or nine?
DARPA’s own published version has eight. A “nine questions” framing shows up in some secondary proposal-writing guides that split “What are the risks and the payoffs?” into two separate questions, since it is really asking two things at once. Both framings cover exactly the same ground — treat the split as a matter of how you organize your answer, not as an official ninth question DARPA added.
Is the Heilmeier Catechism only for DARPA proposals?
No. It was designed for DARPA’s specific culture of funding high-risk, high-payoff research, but the underlying discipline — state the goal in plain language, know the current state of the art, explain what’s genuinely new, and define how success will be measured — applies to any research proposal, and it is taught well outside DARPA for exactly that reason.
Should I answer the questions in the actual proposal text?
Not necessarily verbatim. Most solicitations have their own required sections and page limits. The value of the catechism is as a drafting exercise done before or alongside writing the formal proposal, so that the answers to “who cares” and “how will you know it worked” are baked into the narrative rather than bolted on afterward.
How is this different from a project’s aims and significance sections?
It overlaps heavily with them, but it is more adversarial by design — the phrasing (“who cares,” “what are the limits of current practice”) is meant to be blunt enough that a proposal team can’t dodge the hard question by restating their objective in more technical language.
