NSF — SBIR/STTR competition data

Phase I selection rate — out of everyone who applied, the percent that won. This one is reported by the SBA in its annual reports; we don't compute it.

Phase I→II conversion — out of the companies that already won a Phase I, the percent that later won the bigger Phase II. We compute this from linked award records.

Worked example. NSF's 10.8% selection and 31% conversion means about 11 in 100 applicants win a Phase I, and 31% of those go on to win a Phase II — so roughly 3 in 100 applicants end up with a Phase II.

The two awards are very different sizes. Phase I is the small proof-of-concept award; Phase II is the larger build award (at NSF, a median $305k against $1.2M in FY2025). Winning a Phase I is the cheap half of the story.

Quarterly-ish, and only when something's actually new — a new SBA report year, a new stat, a dataset update. The weekly refresh alone never triggers an email. No account, one-click unsubscribe, and every number on this page stays free and open either way.

Sourced vs computed: Phase I/II selection rates (applications → awards) come from the SBA's SBIR/STTR Annual Report — FY2022 is the latest published edition, and SBA itself calls the rates estimates because proposal cohorts span fiscal years. Phase I→II conversion and first-timer shares are computed by us from sbir.gov's public award records. Linked by firm + agency + topic code where topic codes carry (suffix-tolerant for DOE's phase renumbering), with a title-similarity fallback elsewhere. 70% of links match on topic, 30% on title; each pair's match_tier is in the downloadable CSV, so you can recompute under stricter rules. Firms are identified by federal UEI where available (99.5% of 2025 awards, down to 54% for 2001), with a normalized-name fallback for older records.

Fast-track and Direct-to-Phase-II: no agency publishes a fast-track flag, so we derive it — HHS/NIH: a Phase I record carrying a Phase II activity code (R44/R42/U44) plus its linked Phase II; NSF: the 2024 Fast-Track pilot solicitation (NSF 24-582). Where fast-track exists, conversion is shown both organic (fast-tracks excluded) and all-inclusive, because a fast-track Phase II was committed at award time, not competed. DoD's Direct-to-Phase-II share is different in kind: there is no marker at all, so our number is an inference (Phase IIs with no same-firm DoD Phase I in the prior 3 years), always labeled an estimate and kept visually separate from measured stats.

What nobody publishes: per-topic application counts are published by no one — not the agencies, and not the $500–$42,000-a-year intelligence platforms either; the data simply isn't public. Any per-topic number on this site is award-side only (who won, how often winners repeat), never your odds of selection.

Sources: sbir.gov award data · SBA FY22 Annual Report. Free to use; verify anything critical against the originals.