--- name: consumer-growth-model-and-loop-map description: Use before committing to or materially expanding any Consumer AI Factory venture. Defines the product's acquisition, activation, retention, referral, revenue, and AI-personalization loops so teams do not build useful demos without a growth/retention machine. version: 1.0.0 author: Hermes Agent license: MIT metadata: hermes: tags: [consumer-ai, consumer-software, growth-loops, activation, retention, distribution, monetization] related_skills: [consumer-ai-factory-opportunity-evaluation, consumer-ux-director, consumer-ai-factory-distribution-learning, posthog-venture-telemetry] --- # Consumer Growth Model and Loop Map ## Overview Consumer software rarely works as a linear path of “build useful tool → users arrive → revenue happens.” Strong consumer products are usually built around loops: usage produces the next session, next user, next share, next data asset, or next payment moment. For Consumer AI Factory ventures, this skill prevents the common trap: ```text impressive AI demo + weak habit + weak distribution + no durable loop ``` Use this skill to force a venture’s growth thesis before overbuilding product. The output should be a Growth Loop Canvas that makes the venture’s activation, retention, acquisition, sharing, monetization, and kill criteria explicit. ## When to Use Use before: - Starting a new Consumer AI venture or demand probe. - Deciding whether to build web, mobile, or both. - Expanding from prototype to MVP. - Adding major features before activation/retention is proven. - Reviewing why a product has traffic but weak usage. - Choosing between creator-led, SEO, social, referral, paid, App Store, or community distribution. - Running a weekly venture decision review where the question is continue/revise/park/kill. Do not use as a generic brainstorm. This is a decision artifact: it must produce assumptions, metrics, thresholds, and next experiments. ## Core Principle Every venture should define at least these loops: ```text Acquisition loop: how one action creates the next qualified visitor Activation loop: how a new user reaches the first magic moment Retention loop: why the user comes back naturally Personalization loop: what accumulates and makes the AI better Share/referral loop: what travels outside the product Revenue loop: when willingness to pay is highest ``` If a product has no credible retention loop, it must have a strong sharing/acquisition loop. If it has no credible sharing loop, it must have strong recurring utility. If it has neither, it is likely a demo, not a durable consumer product. ## Consumer Software Laws to Apply ### 1. Loops beat funnels Do not stop at: ```text traffic → signup → activation ``` Map what happens after value is delivered: ```text user gets artifact/result → saves/shares/invites/returns → product gains data/content/trust → next user/session becomes more likely ``` Common loop types: - Artifact sharing loop: generated output is shared, viewers click/remix/create. - Collaboration loop: user invites partner/friend/team/family into shared context. - Content/SEO loop: UGC/templates/tools become indexable and acquire users. - Personalization loop: history/preferences/context improve AI value and switching costs. - Status/identity loop: output helps user perform taste, competence, humor, attractiveness, care, or belonging. - Lifecycle loop: scheduled reminders/digests/check-ins bring users back to unfinished value. ### 2. Activation must be a behavior, not a pageview Bad activation metric: ```text landing_viewed signup_created button_clicked ``` Better activation metrics: ```text first_successful_generation real_context_submitted artifact_saved_or_shared third_conversation_turn partner_invited personalization_source_connected second_session_within_7d ``` For every venture, define the one behavior that predicts retention and decision quality. ### 3. Retention comes from accumulated value Look for one or more: - Stored value: history, projects, memories, relationship maps, saved outputs. - Scheduled value: weekly ritual, reminders, check-ins, digests, streaks. - Social value: partner/friend/community/accountability loop. - Improving value: AI learns preferences, context, taste, constraints, goals. - Emotional value: companionship, confidence, identity, relief, entertainment. A one-off AI answer is weak unless the artifact spreads or the situation naturally recurs. ### 4. Distribution must be designed into the product Do not treat distribution as a separate chore after product design. Ask: - What object leaves the app? - Why would the user share/invite? - What does the recipient see/do next? - Does the product generate indexable or creator-demo-friendly material? - Which channel’s native format does the product fit? - What attribution/watermark/linkback is appropriate without hurting trust? ### 4b. Qualified traffic is a prerequisite for product conclusions If a venture is not receiving reasonable qualified traffic, do not over-interpret activation, retention, or product-quality metrics. The first bottleneck may be distribution, not UX. Use this gate before expanding product work: ```text Where do users already gather? What pain phrase makes them stop? What channel-native artifact earns the click? What channel can produce 100-300 qualified visits/week? What do we learn if it fails? ``` Suggested early-stage thresholds: - Minimum: 100 qualified visits/week to a probe. - Good: 300-500 qualified visits/week. - Strong: 1,000+ qualified visits/week with nonzero activation. Rule: if traffic is zero, do not conclude the product is weak or strong. First create a traffic acquisition thesis and a small controlled sprint. ### 5. Monetization should follow the value moment Consumer AI has real inference costs, but paywalling before the magic moment usually kills learning. Map: - What free action proves the magic? - Which expensive actions need limits/credits? - What saved/exported/continued/partner/premium moment has highest willingness to pay? - Does subscription fit natural recurrence? - Does usage pricing fit output value? - What model/caching/template strategy controls cost? ### Web vs Mobile Rule Default factory sequence: ```text web probe → responsive/mobile-first web MVP → PWA/home-screen tests → native only if the loop requires native behavior ``` Core rule: ```text Become phone-native in behavior before becoming native in code. ``` ### Use web first when testing - Pain and positioning. - Landing conversion. - First input friction. - First output value. - Creator/community/Reddit/email distribution. - SEO/search intent. - Pricing interest. - Fast copy/UX/CTA iteration. - No-install partner/invite links. Web is the factory’s validation weapon. ### Make the web product mobile-first when the moment happens on a phone Before native, test cheap app-like behavior: - Phone-first layout and real-device QA. - Text/screenshot/upload/share flows. - Optional PWA installability and home-screen prompt after a high-value moment. - Discreet email/SMS/in-app/PWA reminders. - Web push only after reminder interest is proven. - Device/browser/mic-permission/import/invite/reminder instrumentation. ### Move to native when testing or scaling - Daily/weekly habit. - Push notifications proven to lift retention. - Camera/photo/audio capture beyond what mobile web can reliably support. - OS share sheet / “send this chat/screenshot to the app.” - Home-screen/lock-screen presence and widgets. - Intimate/private repeated use with local/biometric privacy needs. - Contacts/social graph. - App Store search/ASO with proven keyword demand and review plan. - Native subscription/paywall flow. - On-the-go moment capture where browser friction causes measurable abandonment. Mobile is the retention/distribution weapon, not proof that the product is real. ### Mobile too early is a smell when - Activation is not proven on web. - The main unknown is whether users will submit real context. - Traffic is too low to interpret mobile behavior. - The team wants polish/status instead of learning velocity. - Distribution is still manual/community/creator-link based. - No-install partner invites have not been tested. - The product lacks a clear notification/habit/use-frequency reason. - PWA/home-screen prompt uptake is weak after high-value moments. For a concrete HappyCouple / Emma example with competitor evidence, internal telemetry, thresholds, and experiments, see `references/mobile-app-decision-research-happycouple-2026-05-29.md`. ## Growth Loop Canvas Produce this artifact for every serious venture or major pivot. ```markdown # Growth Loop Canvas — [Venture] ## Product thesis - User: - Pain/desire moment: - Product artifact: - Why AI matters: ## North Star - North Star metric: - Why it represents delivered value: - Current baseline: - Target threshold: ## Activation - Activation event: - Time-to-value target: - First-session path: - Instrumentation events: - Kill/revise threshold: ## Retention - Natural use frequency: - Stored value: - Scheduled value: - Social/emotional value: - D1/D7/D30 metrics: - Kill/revise threshold: ## Acquisition loop - Primary loop: - Source channel: - User action that creates next user/visitor: - Loop input: - Loop output: - Decay point: - First experiment: ## Share/referral loop - Share object or invite object: - Sender motivation: - Recipient motivation: - Recipient activation path: - Instrumentation: ## Personalization/data loop - Data/context accumulated: - How it improves output: - User-visible proof of improvement: - Trust/privacy risk: ## Revenue loop - Monetization moment: - Pricing model: - Free limit: - Cost-control approach: - Revenue metric: ## Platform posture - Web/mobile/default platform: - Why: - Trigger to move platform: ## Next 3 experiments 1. Hypothesis: - Test: - Metric: - Decision rule: 2. Hypothesis: - Test: - Metric: - Decision rule: 3. Hypothesis: - Test: - Metric: - Decision rule: ## What not to build yet - - - ``` ## HappyCouple Example Do not ask only: ```text Is HappyCouple useful? ``` Ask which loop is real: - Creator-led trust loop: specialists/creators send qualified users because the artifact helps their audience. - Partner invite loop: one partner starts, invites the other, shared context increases value. - Shareable relationship artifact loop: user shares a script/report/check-in outcome privately or publicly. - SEO intent loop: high-intent relationship questions land on useful artifact pages. - Recurring check-in loop: weekly relationship ritual creates retention and memory. Near-term platform posture: ```text Keep HappyCouple web-first until strangers submit real context and find the output valuable. Design the web MVP mobile-responsive, but do not build native mobile until retention/partner/voice/check-in behavior proves mobile-native demand. ``` Activation candidate: ```text user shares real relationship context → receives useful relationship artifact/script/report → saves, shares, returns, or invites partner ``` Do-not-build constraints before evidence: - Do not build native mobile just to look more real. - Do not add broad relationship dashboards before first-context submission works. - Do not scale paid acquisition until activation has signal. - Do not optimize long-term memory if users do not finish the first relationship artifact. ## Common Pitfalls 0. **Mistaking content consumption for a product loop.** For anti-doomscroll / anti-internet products, a feed that gets views but does not create real-world action is part of the problem. Social clips should route to an activation behavior such as starting/completing a rep, saving a challenge, reflecting with AI, or adding proof to a scoreboard. Track rep starts/completions, not only views and clicks. 1. **Mistaking a funnel for a loop.** A landing page and signup form are not a growth model. Identify what creates the next user/session. 2. **Calling pageviews activation.** Activation is a behavior that predicts value and retention. 3. **Overbuilding mobile too early.** Mobile is valuable when the loop needs native surfaces; it does not fix weak demand or weak activation. 4. **Ignoring cost loops in AI products.** Free magic must be bounded by credits, cheap models, caching, templates, or paywall moments. 5. **Adding features without changing loop mechanics.** A feature is only strategically important if it improves activation, retention, acquisition, referral, personalization, or monetization. 6. **Overgeneralizing distribution.** “Creators,” “Reddit,” or “SEO” is not a loop. Specify the exact action, artifact, recipient, and conversion path. 7. **No kill threshold.** Every loop assumption needs a measurable threshold and a decision deadline. ## Verification Checklist - [ ] Growth Loop Canvas completed. - [ ] North Star metric is behavior-based, not vanity-based. - [ ] Activation event is specific and instrumentable. - [ ] Retention loop explains why the user returns. - [ ] Acquisition/share loop explains how usage creates more usage or visitors. - [ ] Personalization/data loop explains why AI value improves over time. - [ ] Revenue loop maps to a high-willingness-to-pay moment. - [ ] Web vs mobile posture is explicit. - [ ] Three next experiments have metrics and decision rules. - [ ] “What not to build yet” is listed. ## References See `references/consumer-software-loop-research-2026-05-16.md` for the research synthesis and source list that motivated this skill.