Referrals
Creates, optimizes, or analyzes a referral program, affiliate program, or word-of-mouth strategy.
Growth Engineeringgrowth-engineeringstandard
by AgentVoltv1.0.0Published 1y ago1 install
Free to sign up · every skill included with AgentVolt Pro
Before & after
Same prompt, same agent — the only difference is the skill.
Prompt“Design a referral incentive that won’t get abused.”
Beforegeneric AI agent
Designing a referral program that resists abuse requires some thought. Here are strategies: - **Reward on a qualifying action**, not just signup (e.g., first purchase) - **Set limits** on rewards per user - **Verify referrals** to catch fake accounts - **Use double-sided incentives** to align both parties - **Monitor for fraud patterns** like self-referrals The core tension is making the incentive attractive enough to drive referrals while not being so generous that it invites gaming. Credit or account-based rewards tend to be less abusable than cash, since they keep value inside your product. What kind of product are you designing this for?
Afterwith Referrals
Abuse-resistance is a mechanics problem, and it's mostly decided by ONE
choice: when the reward fires. Get that right and most fraud vectors
close on their own.
THE CORE DECISION — reward trigger
On signup → maximum fraud. You're paying for account creation,
which costs a referrer nothing to fake.
On first payment → most fraud dies here. A fake referral now has to
put real money in, which defeats the point.
On 2nd payment / retention → near-zero fraud, some referral lag.
→ Recommend: gate on first successful payment. Best fraud-to-friction
ratio for most products.
REWARD TYPE — credit, not cash
Cash attracts people who want cash and will game for it. Account credit
keeps value inside the product and is worthless to a fraud ring — you've
removed the motive, not just added a check.
THE THREE CHECKS THAT CATCH THE REST
1. Self-referral block: same payment fingerprint OR same email domain
+ same device/IP between referrer and referee.
2. Velocity cap: N rewarded referrals per user per month. Above the
cap it's affiliate behavior — route it to a real affiliate program
with a contract, don't reward it silently.
3. Clawback window: reward reverses if the referred payment refunds
inside 30 days. Closes the "pay, collect, refund" loop.
DOUBLE-SIDED, INTENTIONALLY
Both sides get value — but the referrer's reward gates on the referee's
real action. This aligns incentives: the referrer is motivated to send
people who'll actually convert, not just click.
THE ABUSE YOU CAN'T FULLY DESIGN OUT
Genuine-looking low-quality referrals — real people, real payment, then
churn immediately. The retention-gated trigger and clawback are your
only defense; monitor referred-cohort churn separately and if it runs
hot, move the trigger later.
What's the product and price point? That sets whether "first payment" or
"retention" is the right trigger.About this skill
name: referrals description: Use when Creates, optimizes, or analyzes a referral program, affiliate program, or word-of-mouth strategy.
Referrals
Designs the mechanics and incentives of a referral or affiliate program.
What you get
- A referral program structure with incentive design.
Customize your output
- Reward type
- referral trigger point.
Example output
A referral program design brief.
Best for
Products relying on word-of-mouth growth.
SKILL.md preview
SKILL.md
---
name: referrals
description: Use this skill when creating, optimizing, or analyzing a referral program, affiliate program, or word-of-mouth growth strategy.
version: 1.0.0
category: Growth Engineering
author: AgentVolt
license: proprietary
tags:
- growth-engineering
- standard
---
# Referrals
Designs the mechanics and incentive structure of a referral or affiliate program, or diagnoses why an existing one isn't producing referrals.
## When to use
… (sign up to view the full skill)Sign up to view, copy, and install the full skill