Growth Loops

Funnels are linear: spend → acquisition → retention → churn (capital-intensive). Loops are circular: user action → network effect → new user → action (capital-efficient, compounds). Pinterest: user saves pins (output) → appear in feed → new user discovers → saves → feeds others. Slack: user invites team (output) → workspace grows → attracts adjacent team. Notion: user builds database (output) → shareable link → new user forks → creates more. Funnels optimize CAC (cost per acquisition); loops optimize K-factor (new users per existing user). Most successful products run multiple loop types simultaneously.

Viral loops (K-factor): user action directly invites friends (Slack, Dropbox, referral bonuses). Speed: minutes-to-hours, but fake virality if forced invites. Content loops (SEO): user creates content → Google indexes → attracts new users → creates more content (Pinterest, YouTube, TikTok, Notion). Speed: weeks-to-months, extremely scalable. Paid loops (revenue funds acquisition): user pays → revenue funds ads → acquisition → some pay → repeats. Speed: days-to-weeks, predictable but linear. Sales loops (enterprise deals): customer success → case study → new prospect → deal. Speed: months, lower velocity but high LTV. Most efficient: content > viral > paid > sales.

K-factor (viral coefficient) = (# invited × conversion rate). K > 1 = viral (net growth per user), K < 1 = decay (needs paid/content). Cycle time = days from action to new user signup. Short cycle (hours-days) = 2x/week measurement possible; long cycle (months) = requires email/tracking pixel. Loop ROI = revenue per loop ÷ cost to optimize loop. Example: content loop at 100k organic/mo at $0 CAC (owned channel) vs paid loop at 50k/mo at $30 CAC = choose content. Measure separately: acquisition loop (funnel to first action), activation loop (first action → daily active), monetization loop (revenue generation → reinvestment), retention loop (engagement → return). Immature programs track vanity; mature programs track cycle-time, K-factor, and contribution to NRR.

TikTok: algorithm is the loop (user watches → engagement signal → more relevant videos → watch-time magic). As a creator/brand, you don't control the loop; you feed it (post consistently, hook in 0.5s, CTR/retention to feed). Creator funding via TikTok Creator Fund or Shop ads closes a monetization loop. Notion: collaborative docs = content loop (user builds → shares link → friend adds pages → shows in social feed → discovered by others). Notion templates marketplace extends this (creator publishes → user forks → customizes → shares). Pinterest: save/repin behavior = content loop (user saves DIY idea → appears in saved collections → friend discovers → creates new pins from idea). Algorithm amplifies unique pins (repins → impressions → source traffic). Best strategy: create loops that match platform mechanics (don't force Slack-style invites on TikTok; lean into algorithm/content discovery instead).

Loops break when any step deteriorates: (1) output drops, users stop taking action (feature regression, fatigue, algorithm change). Debug: compare last month's action rate, check if feature or external factor changed. (2) Virality drops, invitation acceptance or signup conversion falls. Debug: A/B test invite copy/timing/channel. (3) Cycle time inflates, new users signup slower. Debug: sample of recent signups; interview to understand journey. (4) External loop killed, platform changes algorithm (TikTok restriction), API deprecation (Twitter removed external sharing), competitor saturation. Debug: cohort analysis to see when drop started. (5) Saturation, all your users who can viral already have. Debug: model loop asymptotically; need new loops or paid acceleration. Example: WhatsApp's viral loop worked until market saturated (2015); they added bot integrations to start a new ecosystem loop. Fix first loop to maintain velocity; build new loop in parallel.

AI changes where loops sit in the stack. (1) Content recommendation (TikTok, Netflix, Spotify): algorithm learns preferences → personalized feed → watch-time compounds → user status (influencer, power user) → network effect. AI tunes K-factor automatically. (2) AI-generated content: users prompt AI → content generated → shareable → attracts audience → new users prompt → content loop accelerates (ChatGPT plugins, Claude, Gemini). Challenge: quality + copyright complexity. (3) Agentic loops: AI agents act on behalf of users (customer support bot → resolves tickets → increases CSAT → reduces churn → improves NRR → compounds). (4) Personalization-as-moat: Amazon (user buys → recommendations improve → buy-through-rate up → network more valuable → competitive advantage). In 2026, loops powered by AI recommendation + agentic behavior outpace manual loops. Strategy: measure loop K-factor at the model level (does Claude plugin loop drive activation vs Slack integration loop). Most competitive advantage goes to teams optimizing AI loop feedback speed (daily model refreshes vs weekly metrics reviews).

Build loops first if: (1) product has inherent shareability (calendar invite, collab doc, invite-only beta), (2) target market is networked (enterprise, B2B SaaS, gaming, social apps), (3) you have <$100k/mo budget (loops = capital-efficient). Prioritize paid if: (1) sales cycle is long (enterprise, real estate, SAAS for SMB), (2) loops don't fit product (B2B2C with fragmented networks), (3) you need predictable, linear growth (acquisition-dependent business model). Prioritize SEO/content if: (1) buyer journey = research-heavy (health, finance, career), (2) content production cost is low for your team, (3) loop virality is weak but intent-to-action is high (JobCannon test → blog drives loop; blog → long tail keyword → person → test). Rule: design the primary loop (fastest K-factor + shortest cycle time) in months 1-3. Run secondary loop (next-best K-factor) months 4-6. Add paid as multiplier (capital lets you scale proven loops) months 6+. Mature programs run all three; new companies pick one to master.