referral-program-design

Installation
SKILL.md

Referral Program Design

Design a referral program that asks the right customers at the right moment, rewards them in a way that does not buy the wrong customers, and costs less per customer than the channel it replaces.

When to use

The user wants to start a referral or advocacy program, size a reward, decide between one- and two-sided incentives, or check whether an existing program pays for itself. For the general loop math across referral, content and paid loops, use aaj-marketing-loops; this skill is the referral program itself.

Before you start

  1. Read the brand/product context first. Pull ICP, pricing, buyer, and whether customers are healthcare providers from .agents/product-marketing.md (or .agents/aaj-brand.md). If none exists, ask.
  2. Gather inputs: active customers; the signals that mark an engaged customer (NPS promoters, usage depth, renewals, a milestone reached); the share of customers who meet them; LTV on gross margin; the fully loaded CAC of the channel referred customers would otherwise come from; gross margin per customer per month.
  3. Separate data from guesses. Ask-to-referrer and referral-to-customer rates are usually unknown before launch. Mark them as assumptions and treat the first period as the test that replaces them.
  4. Confirm the objective: more customers at a lower CAC, references for sales, or both. In B2B, references and introductions often matter more than a cash reward.

Method

Diagnose whether referral fits: customers can show a result they are proud of, they know people with the same problem, and those people trust them. If any of the three is missing, fix the product or the segment before building a program.

Installs
3
GitHub Stars
1
First Seen
Oct 12, 2026