Searching SBIR topics by keyword will find you the wrong ones. Topics are written by technical staff describing a capability the government is missing, not by marketers describing a product category, so the words that matter are often buried in the background paragraph rather than the title. The skill worth learning is reading a topic for the gap underneath it.
Where to look, and how often
Start with the government-wide topic search, which aggregates open topics across all participating agencies and lets you filter by agency and status. Then go to the two or three agencies whose mission plausibly covers your technology and read their topic lists directly — agency sites carry context, past awards and program office notes that an aggregator strips out. Reading past awards under a topic number is one of the fastest ways to learn what that office considers a good answer.
Do this on a schedule, not when you happen to remember. Agencies release on their own cycles and a topic you find two days before the close date is a topic you have already lost. If you have not narrowed down agencies yet, our comparison of which federal agency to apply to is the place to start, and the SBIR and STTR library covers the surrounding mechanics.
Reading a topic for the gap, not the keywords
Ask four questions of any topic that catches your eye. What can the government not currently do? Why have existing vendors failed to solve it — cost, size, power, durability, accuracy, or something regulatory? What would the office do with the result operationally? And what does the Phase I deliverable list imply about how far along they expect you to already be?
Those answers often reveal that a topic using none of your vocabulary is a strong match, and that a topic full of your keywords is really asking for something adjacent. A topic about "field-deployable detection" may be, underneath, a topic about power budget and false alarm rates. Write to the gap.
Honest fit versus expensive stretching
A genuine fit means you can address every stated requirement without inventing a new capability, your team can show relevant prior work, and the Phase I objectives are achievable in the stated period with the stated funding. A stretch shows up as vague language, a proposal that spends its pages on what you have already built, and requirements you quietly skip.
The cost of forcing a poor fit is not just the loss. It is four to eight weeks of founder time, a debrief that tells you what you already suspected, and a cycle you could have spent preparing for a topic that actually matched. Base rates are unforgiving even for good fits — see our piece on SBIR success rates — and a stretch sits well below them. If your technology is software rather than hardware, the fit question has its own shape, covered in can a software product win an SBIR.
Projects House helps founders judge, honestly, whether a topic is buildable with their technology in a Phase I period. Send us a topic number through the contact form.