viral_loop_design v1.0.0

name: viral_loop_design description: Design a viral growth loop for a product: pick the right loop type, model the viral coefficient with honest math, choose share triggers, and spec referral mechanics. Triggers on "viral loop", "growth loop", "referral program", "K-factor", "make this go viral", "network effects", "share triggers".

Viral Loop Design

Purpose

Turn a product into its own acquisition channel. Given a product and an audience, this skill designs a concrete viral loop: who shares, what makes them share, what the invite looks like, what the reward is, and what the numbers say. It runs on honest math, not growth-hype.

Workflow

1. Pick the loop type. Four options (see references/loop-patterns.md):

• Inherent: the product gets better when friends join (messaging, collaboration).
• Incentivized: you pay for referrals (credits, cash, storage, unlocks).
• Usage exposure: non-users see the product in use (shared docs, watermarks, badges).
• Content: users produce shareable outputs (embeds, generated media, portfolios). Pick the one the product already leans toward. Do not force an incentivized loop onto a product with no natural sharing moment.

1. Model the viral math. K = i × c, where i is average invites sent per user and c is the invite-to-active-user conversion rate.

• Estimate i from behavior that already happens: how many people already share the product unprompted? That is the floor. Do not invent it.
• Estimate c with a landing page or waitlist test before building anything.
• Use scripts/kfactor.py to compute K and project cohort growth.
• Set the target as share of new signups from referrals, plus a K tracking number. Plan for decay: K above 1 is rare and fades as the audience saturates; 0.2-0.4 is a healthy contribution for most products.

1. Choose 1-2 share triggers (STEPPS). Jonah Berger's six drivers of sharing: Social currency, Triggers, Emotion, Public, Practical value, Stories (see reference). Do not try to hit all six; the best loops nail one or two. Name the exact trigger and where it fires in the product flow.

1. Design the incentives (incentivized loops only).

• Double-sided by default: both referrer and referee get something.
• Match reward to product type: account credit (subscription), discount (ecommerce), feature unlock (freemium), cash or storage (marketplaces/utilities).
• Size against paid CAC: total reward cost per referred user must stay well under paid CAC.
• Pay on activation, not signup. This filters junk referrals and self-dealing.
• Ask at high-satisfaction moments: right after a win, a saved result, or a 9-10 NPS/CSAT score. Never during onboarding friction or error states.

1. Spec the mechanics.

• The share surface: one-click, pre-filled message, contact import where it fits.
• The recipient landing: what the invitee sees and why they convert.
• Attribution: unique links or codes, fraud checks (rate limits, activation requirements, self-referral flags).
• Instrumentation: invites sent, acceptance rate, signup conversion, activation conversion, viral cycle time (days from signup to first invite).

1. Ship an experiment backlog. Ordered list of what to test first: copy variants, reward sizes, placement of the ask, landing page. Each experiment names the metric it moves (i or c).

Output Contract

A loop spec that includes:

• Loop type, target user, and the exact sharing moment.
• Viral model: i, c, K, cycle time, target share of new signups from referrals, decay assumption.
• Incentive design: reward type and amount, double-sided structure, CAC comparison, pay-on-activation rule.
• Share surface and recipient landing description.
• Fraud and attribution plan.
• Experiment backlog with metric targets.

Operating Rules

1. Honest math first. If the estimated K is below 0.2, say the loop will not carry growth and recommend where the product actually acquires users. Never inflate i or c to make the math work.
2. No invented numbers. Benchmarks in the reference are labeled as what they are (heuristics, reported cases). Estimates for the user's product come from their data or a test, never from memory.
3. Every factual claim about a case study (Dropbox, Hotmail, WhatsApp, Blendtec, Kit Kat) must match the verified notes in references/loop-patterns.md.
4. Valuable virality check. The product must be integral to whatever spreads. If someone could retell the story without mentioning the product, the virality is not valuable. Redesign until the product is essential to the story.
5. Fraud gates are not optional on incentivized loops: activation requirement, rate limits, self-referral detection.
6. Style: no em dashes, ever. No AI-template phrasing. Write like a person.