The workstream behind end-to-end onboarding test runs — how a real customer travels from "finds us" to "in the product." Internal — pre-launch working drafts, real companies named. Bands and statuses are not final; confirm before external use.
The full matrix the team makes product decisions against: one engine (role → what Perch does → their $ lever → the next question), then six anchor journeys — GC and sub, small to enterprise — each bent by who lands. Plus the role-bend (estimator vs owner vs field) and the trade-bend (Electrical / Concrete / Drywall — counts vs volume vs area), with the other 31 trades following the same spine.
The reimagined flow: email is the only ask, sub-agents research the company on capture, and onboarding becomes confirm-don't-ask + an instant role-specific value demo. Two tracks (self-serve = the agent sells; white-glove = research arms a human), four concrete persona journeys, racing three clocks: time-to-info, time-to-value, time-to-check.
The mental model behind it all: three related-but-distinct layers (find us → set up → use it), Brock's point that the last 4–5 steps are really initialization (where stickiness is made), the role overlay (each role uses Perch differently), and the sub-agent research that fires on first contact to pre-fill and steer onboarding.
The strategy-room whiteboard, mapped: the customer universe (GC × 1, Sub × 34 trades) with revenue bands that set the journey, the six acquisition channels, the two journeys (self-serve vs white-glove, gated by ACV), the 8-step onboarding flow, and the seven conflicts it surfaces against the test-client set.
Real, verified Utah companies → exemplars we model mock-up customers on for end-to-end onboarding/sales test runs. v0 format-check batch: 3 GCs (S/M/L) + 5 trades × med/large + 1 self-serve micro. Scales to the full 34 trades once the bands and field set are locked.