Growth and monetization
Referral programs
A designed viral loop. Ask activated users, reward in-product currency, measure retention — and budget for fraud.
Referrals are a designed viral loop: explicit incentives for inviter and/or invitee. Dropbox's classic two-sided reward (extra space for both) is the canonical case. The reward was in-product currency aligned with the value metric, the ask came after the user felt value, and the loop was easy to understand.
Collaboration virality (must invite to get value) has better intent and lower fraud but only fits multiplayer products. Referral bonuses work for single-player tools but can attract junk signups.
Concept
A referral program is an incentive layered on top of a K-factor loop. Measure referred users' activation and retention vs other channels — not just referral signup count.
Walking through the checklist
- Who to ask: activated users, not brand-new signups.
- What to reward: something that increases usage (credits, storage, seats) over pure cash when possible — cash attracts referral professionals and low-quality signups.
- Two-sided vs one-sided: two-sided usually converts better; one-sided can be enough if the product is already highly shareable.
- Fraud and gaming: self-referrals, disposable emails, incentive stacking. Budget engineering time for this or the loop's unit economics lie.
- Measurement: activation and retention of referred users, plus K on activated users.
Trade-off
Prefer natural collaboration hooks when the job is shared. Add referrals when the product is valuable alone and you need an extra push. Fake gates ("invite 5 friends to unlock") on a single-player tool inflate i, crush c, and teach users you are extractive.
Engineers own the mechanics: signed invite tokens with expiry, rate limits to blunt referral fraud, Open Graph previews that raise invite CTR, and idempotent referral credit ledgers so double-grants do not bankrupt a Dropbox-style reward.
Instrument the edges:
activated_users
→ % who hit invite affordance
→ invites_sent / activated_user (i)
→ invite_link_ctr
→ signup_rate from invite (feeds c)
→ activation_rate of invitees
→ their invites… (next turn)In interviews, draw the loop, label the rates, estimate K and cycle time with rough numbers, and say what you would instrument first.
Common mistakes
- Asking at signup. The user has not felt value; invites will be junk.
- Cash to anyone. Professionals and throwaway accounts show up.
- Counting signups, not retained referred users. Week-1 K can hide week-4 emptiness.
- No fraud budget. Self-referrals make the unit economics a lie.
Check yourself
When do you show the referral ask?
After activation — once the user has felt value — not on the signup confirmation screen.
Why did Dropbox give space, not cash?
Space is in-product currency aligned with the value metric. It increases usage instead of attracting bounty hunters.
Key takeaways
- Ask activated users. Reward in-product currency when you can. Prefer two-sided rewards.
- Measure referred activation and retention, not signup count.
- Budget for fraud: tokens, rate limits, idempotent credit ledgers.
Further learning
Back to the map in growth loops.
Related
Growth loops and acquisition
Funnels vs loops, virality math, content and paid loops, and referral programs.
K-factor
K = invites per user × conversion. Below 1 still cuts CAC. Cycle time sets how fast generations stack.
Activation and onboarding
Aha moments, time-to-value and onboarding flows that turn signups into active users.