SAM.gov models federal contracting as flat tables of opportunities. We don't. Bridger models it as a directed graph: agencies buy from primes, primes sub to allied vendors, vendors hold past performance with previous primes and agencies. The Past Performance Graph is the queryable core of that model.
Past performance entries live in the PastPerformance Prisma model with explicit edges to Agency, Prime/Sub role, NAICS, and CPARS rating. The graph traversal is implemented as a recursive CTE on Postgres — fast enough that a contractor can query “past performance relevant to this opportunity” in under 300ms even with thousands of entries.
The opportunity-matching engine uses past-performance edges as one of three primary signals (alongside NAICS overlap and tech-area overlap). Without past performance, you’re a stranger; with even one relevant edge, you’re a known counterparty.
Foreign vendors entering the US market who have substantial allied past performance but no US-side history yet. The graph shows you which of your existing edges are actually transferable.
Mid-tier US contractors in subcontracting roles. Your past performance under prior primes is often invisible to the next prime team — the graph makes it discoverable.