Reviewer feedback on a declined SBIR or STTR proposal is the most valuable document you will get from the agency, and founders routinely waste it by reading it once, angrily, and filing it away. Read it a second time a week later, with a pen, and treat every comment as evidence about how a smart, hurried, non-specialist reader experienced your document.
What the scores and comments actually tell you
Agencies package feedback differently. NIH returns a summary statement with criterion scores and written critiques; NSF returns panel and individual reviews with ratings; DoD components often return far less, sometimes only a short evaluation against the topic criteria. In every format, the numbers and the prose answer different questions. The scores tell you which criterion sank you — significance, innovation, approach, team, commercial potential. The comments tell you why. If your approach score is weak, the fix is in the work plan; if significance is weak, the fix is in the framing of the problem, and no amount of experimental detail will help. It is worth knowing how these panels are structured and scored before you interpret a single line.
Presentation problem or fit problem
Sort each comment into one of two buckets. A presentation problem sounds like "the milestones were not clearly tied to go/no-go criteria" or "the path to a first customer was asserted rather than shown." Those are writing and evidence gaps, and they are fixable in one cycle. A fit problem sounds like "this is incremental engineering rather than research," "the risk being retired is not the risk the topic cares about," or "this appears to be product development the private market would fund." Those say the agency is not the buyer. Piling more detail onto a fit problem is the single most common way founders burn a second cycle.
When two reviewers contradict each other
It happens constantly, and it is not a mistake in the process. One reviewer wanting more preliminary data while another calls the same work overly cautious usually means the proposal did not make its own case clearly enough for either to anchor on. Look for what the contradiction has in common: both reviewers were uncertain about the same section. Rewrite that section so it states its position plainly rather than trying to satisfy both critiques at once. Where a comment is factually wrong — a reviewer missed a paragraph that was there — the honest conclusion is usually that the paragraph was not findable, not that the reviewer was careless.
Remember the panel changes
Reviewers rotate between cycles, so the people reading your next submission may not be the people who wrote these comments. Two consequences follow. First, do not write your next proposal as a rebuttal aimed at named criticisms; write it as a standalone document that happens to have absorbed them. Second, a comment that only one reviewer raised, and that the panel did not converge on, deserves less weight than one that shows up in multiple reviews. Comments that repeat are the signal. Given the base rates in these programs, a near-miss score with consistent minor critiques is a good sign, not a bad one, and a call to the program officer can tell you whether they agree.
Projects House helps founders close the technical gaps reviewers point at, from feasibility testing to working prototypes. Send us your critique summary through the contact form.