the-referral-architectlisted

> **Choose the ask moment from measured satisfaction, not from lifecycle stage.** "After the second > purchase" is a proxy. The strongest referral moments are a **resolved support issue** — where the > customer just experienced the company being good at something under pressure — and immediately after > a positive survey response. Ask which of those the user can actually detect and trigger on. Where > neither is instrumented, say the stage-based moment is a fallback and name the event that would > replace it, because that is the single highest-leverage change to the programme. > **Boundary:** For a single one-off "ask this happy customer for an intro" message, draft it directly rather than running a full skill. This skill designs the systemic, repeatable referral program, not a one-time favor. ## Context 1. Check for `.agents/product-context.md`: if missing, ask the user to run `/gtm:product-context` first. If the user prefers to proceed without it, ask inline for: what's being sold, the price point, and the business model (subscription, one-time purchase, usage-based). 2. Read `references/referral-incentive-benchmarks.md` for incentive-structure patterns by pricing tier, trigger-timing benchmarks, and anti-abuse patterns by incentive type. ## Inputs 3. Ask: "Who is your advocate pool: which existing customers would realistically refer someone, and at what lifecycle stage are they usually happiest?" 4. Ask: "Has a referral program been tried before? If so, what happened