# Referral Programs

> When to activate: referral program, referral marketing, word of mouth, viral coefficient, refer-a-friend, double-sided rewards, referral funnel, growth loops

- Skill: `mattakushi432/referral-programs` (Agent Skill)
- Install (CLI): `npx skillmds@latest add mattakushi432/referral-programs`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mattakushi432/referral-programs/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Marketing & Growth
- Author: Mattakushi432 (https://skillmd.com/u/mattakushi432)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mattakushi432/referral-programs

---


# Referral Programs

## Core Mechanics

### Referral Program Types

| Type | Reward structure | Best for |
|------|-----------------|---------|
| Double-sided | Both referrer and referee get reward | B2C, SaaS trials |
| One-sided | Only referrer gets reward | High-margin products |
| Tiered | More referrals = better rewards | Communities, loyalty |
| Give-to-get | Referrer gives reward to friend | Gifting, generous brand positioning |
| Social proof | Public leaderboard, recognition only | Community-driven products |

### Reward Design Principles
1. **Aligned with value**: Cash for B2B, credits for SaaS, discount for e-commerce
2. **Immediate enough**: Reward within 24h of conversion, not 30 days
3. **Proportionate**: Reward ≥ 10% of CLV for referrer
4. **Scalable**: Cap rewards or set conditions (e.g., referee must pay)

---

## Viral Coefficient Formula

```
K = i × c

Where:
  i = average invites sent per user
  c = conversion rate of invites to signups

K > 1 = viral growth (each user generates >1 new user)
K = 0.5–1 = significant amplification
K < 0.3 = referral adds noise, not signal
```

**Example:**
- Average user sends 5 invites
- 20% convert to signups
- K = 5 × 0.20 = 1.0 (break-even viral)

To double K: either double invites sent (UX) or double conversion rate (landing page/offer).

---

## Referral Funnel

### Stage 1: Trigger — Make Referring Easy to Think Of
- Show referral CTA at moments of delight (after first value, after milestone)
- Email trigger: "You love [Product]? Your friends will too"
- In-app: after positive NPS or review prompt

### Stage 2: Share — Remove Friction
- One-click link copy
- Pre-filled social share messages
- Email invite with pre-written copy
- WhatsApp / SMS share buttons on mobile

### Stage 3: Land — Convert the Referred Friend
- Dedicated referral landing page (not generic homepage)
- Personalized: "Your friend [Name] invited you to try [Product]"
- Urgency: "Claim your [reward] — offer expires in 72 hours"
- Clear CTA above the fold

### Stage 4: Activate — Confirm the Reward
- Email confirmation: "Your reward is unlocked!"
- Show reward balance in dashboard
- Notify referrer when friend converts: "Sarah just signed up using your link!"

### Stage 5: Loop — Encourage Repeat Referrals
- Progress bar toward next reward tier
- Monthly reminder: "You've earned $X — share again to earn more"
- Leaderboard (optional, if competition is on-brand)

---

## A/B Testing Referral Rewards

### What to Test

| Variable | Option A | Option B |
|----------|----------|----------|
| Reward type | $10 cash | $15 credit |
| Timing | Immediate | After 30 days active |
| Structure | Double-sided | One-sided (referrer only) |
| Framing | "Earn $10" | "Give $10 to a friend" |
| CTA placement | Post-signup | Post-activation (day 7) |

### Significance Threshold
- Minimum 200 shares per variant before concluding
- Use two-proportion z-test for conversion rates
- Primary metric: referred user activation rate (not just signup)

---

## Fraud Prevention

### Common Fraud Types
- Self-referral (user signs up with second email to earn both sides)
- Bulk fake signups from one referrer
- Referral code sharing on coupon sites

### Prevention Measures
- [ ] Require payment method before reward unlocks
- [ ] Flag multiple accounts from same IP / device fingerprint
- [ ] Set minimum activity threshold (e.g., 7 days active, first payment)
- [ ] Review outliers: anyone with >10 referrals in first week
- [ ] Exclude referral reward from first 48 hours of signup
- [ ] One reward per email domain (blocks self-referral via work email variants)

### Fraud Detection SQL (example)
```sql
-- Find suspicious referrers
SELECT referrer_id, COUNT(*) as referrals, 
       COUNT(DISTINCT referee_ip) as unique_ips
FROM referrals
WHERE created_at > NOW() - INTERVAL '7 days'
GROUP BY referrer_id
HAVING COUNT(*) > 10 AND COUNT(DISTINCT referee_ip) < COUNT(*) * 0.5
ORDER BY referrals DESC;
```

---

## Referral Program Optimization Checklist

### Launch Readiness
- [ ] Unique referral link generated per user
- [ ] Referral tracking (UTM + cookie + server-side)
- [ ] Double-spend protection in place
- [ ] Terms & conditions page live
- [ ] Reward fulfillment automated (not manual)
- [ ] Email flows: share prompt, confirmation, reward delivery

### Ongoing Optimization
- [ ] Weekly: review K coefficient trend
- [ ] Weekly: check fraud flags
- [ ] Monthly: A/B test one variable
- [ ] Monthly: email reactivation to non-referrers
- [ ] Quarterly: review reward economics vs CAC

---

## Economics Sanity Check

```
Referral program is worth running if:

  Reward cost < (CLV of referred user) × conversion rate

Example:
  CLV = $500
  Referral conversion rate = 30%
  Expected CLV from referral = $500 × 0.30 = $150

  Double-sided reward = $20 + $20 = $40

  $40 < $150 ✅ Program is profitable

  Also compare: Referral CAC vs. Paid CAC
  If referral CAC = $40 and paid CAC = $120 → 3x more efficient
```

---

## Case Study Patterns

### Dropbox (Give storage, get storage)
- Double-sided: both get 500MB free storage
- Drove 3900% growth in 15 months
- Reward was the product itself → zero marginal cost

### Airbnb (Travel credit)
- Double-sided: $25 travel credit for both parties
- Required first booking to unlock → qualified referrals only
- Fraud layer: payment card required

### Uber (First-ride free)
- Referrer gets $5, referee gets free first ride
- One-sided economic risk but high conversion
- Geo-targeted rewards matched to market

**Pattern:** Most successful programs reward an action beyond signup (first purchase, first key action) to filter for quality.