$ cat projects/fast-path-not-built.md
The Fast Path We Didn't Build
A feasibility study on near-instant screening results, and the case against them.
period: 2025
// 01 problem
Problem
The premise was appealing: could background checks come back in seconds rather than the usual turnaround? Instant results are an obvious differentiator in screening, and the innovation team was asked whether a faster path through the existing flow was achievable.
// 02 approach
Approach
Rather than start building the fast path, we started by measuring where the time actually went — which jurisdictions and which stages accounted for the delay, and which of those a short circuit could realistically remove. Alongside the measurement we put up a rough proof of concept: a GraphQL layer that resolved a query by reaching out to whichever internal subsystems it actually needed, so the shape of the request drove the work rather than a call sequence fixed before we knew what a fast path would have to touch.
// 03 outcome
Outcome
The premise did not survive the measurement. The delay was concentrated in stages a short circuit could not remove, so a fast path would have meant a second pipeline to maintain in exchange for very little a customer would notice. We recommended against it. I include this deliberately: choosing not to build something, and being able to show why, is as much of the job as shipping.