Crucible · Seed-to-Harvest Index v2026.2

Methodology

The Seed-to-Harvest Index measures how much of the follow-on federal contracting won by a program's SBIR firms comes back to the agency that sponsored the SBIR awards.

The index

For each program, every SBIR awardee firm is matched to its later federal prime contracts. Each matched contract dollar is classified as either awarded by the sponsoring agency or by another agency. The index is:

retention = sponsor follow-on $ ÷ (sponsor follow-on $ + other-agency follow-on $)

The denominator is every matched prime-contract dollar won by the program's SBIR firms — summed per firm, so a firm that wins from both its sponsor and other agencies has all of its dollars counted on both sides of the ratio. Programs are ranked by retention; the ranking and shares are versioned (currently v2026.2, generated 2026-08-27) and recomputed whenever the source data is refreshed.

Revisions

• v2026.2 — corrected the retention denominator. v2026.1 summed two per-outcome buckets in which a sponsor-transitioning firm's other-agency dollars were recorded in neither, which understated the denominator and overstated every retention share (for example DoD was published as 99.7% and is 76.2%; NASA as 46.1% and is 10.2%). The published figures now come from each firm's full matched-contract totals, as described above. Any figure cited from v2026.1 should be replaced.
• v2026.1 — initial publication.

Data sources

• SBIR award data: SBIR.gov bulk award downloads, per agency.
• Follow-on contracts: USAspending.gov prime contract awards, matched to SBIR firms by verified recipient-name match.

Data vintages

• DHS — SBIR.gov bulk data last modified 2026-08-01
• DARPA — SBIR.gov bulk data last modified 2026-08-01
• DoD — SBIR.gov bulk data last modified 2026-08-01
• NASA — SBIR.gov bulk data last modified 2026-08-01
• DOE — SBIR.gov bulk data last modified 2026-08-01
• NSF — SBIR.gov bulk data last modified 2026-08-01
• HHS — SBIR.gov bulk data last modified 2026-08-01

Limits

• Association, not causation. A follow-on contract won by an SBIR firm is not necessarily a commercialization of its SBIR work — the index measures where SBIR-seeded firms' later prime contract dollars land, not why.
• Name matching has limits. Firms are joined to USAspending by verified recipient name; firms that renamed, merged, or were acquired can be missed, and the match can under- or over-count at the margins.
• Subcontracts are invisible. Only prime contracts are counted; an SBIR firm working as a subcontractor to a prime does not appear.
• The destination matrix is approximate. It is built from each firm's top follow-on contracts (up to six per firm, by dollars; firms average roughly 23 contracts). Per-program coverage of total matched follow-on dollars: DARPA 48.0%, DoD 53.2%, HHS 45.1%, NASA 45.8%, DHS 49.3%, DOE 48.6%, NSF 53.0%. Retention shares do not use this approximation.
• DARPA is a subset of DoD. The DARPA program rows are also contained in the DoD-wide figures and are shown for comparison only.

Back to the Seed-to-Harvest Index →  ·  Open the interactive pipeline