Honest numbers: Phase I win rates at the major agencies cluster around 15–25%. NIH publishes real data (roughly 15–18% recently). NSF funds maybe a quarter of full proposals — but its mandatory Project Pitch pre-screen turns away about three of four pitches first. DoD varies topic by topic. About 40% of Phase I winners go on to win Phase II. First-timers win every single cycle; the program is designed for them.
Ask reviewers what sinks proposals and you hear the same thing: a giant "TAM" number, no named customers, no competitive analysis, no revenue logic, no letters of support. The fix is unglamorous — talk to ten real potential customers and quote them, name the incumbent you displace, price the product, and get two or three letters. A modest, specific plan beats a grand vague one.
Phase I is officially about establishing feasibility, but proposals that already show early evidence — bench data, a rough prototype, or computational modeling results — read as lower-risk and score better. For hardware concepts, a competent simulation study is often the only preliminary data a pre-funding team can afford: it shows the physics closes, and it shows the team can execute the Phase I work plan.
Reviewers reward preliminary data, and for physics-heavy concepts the cheapest legitimate preliminary data is a well-built simulation: a CFD sweep, a stress margin study, a thermal budget. That's exactly the kind of early work an Ansys evaluation plus an application engineer can support before your funding lands.
See if you qualify for an Ansys eval + engineering support at no cost How access works →At NIH, one resubmission (the "A1") with a point-by-point response to reviewer critiques is the expected path — resubmissions historically fund at meaningfully higher rates than first attempts. At NSF you revise substantively and start again from a Project Pitch. At DoD, topics close and don't recur identically, so you target the next similar topic. Read every critique twice, fix the commercialization section first, and go again.
Also see the two mechanical pitfalls that get proposals returned unread: PI eligibility ambiguity and late registrations — both covered in the registration checklist.
Official sources for this guide: SBIR.gov — evaluation criteria tutorial · NSF merit review · NIH RePORT (success-rate data). Figures change; always confirm on the official page before relying on them.
