app-growth-design 1.0.0

package/skill/references/growth-marketing.md

@@ -0,0 +1,678 @@
+# Growth, Marketing & Positioning
+
+Distilled from 23 of the 30 Tim Gabe notes. Source markers `[NN]` map to the list at the bottom.
+Scope: distribution, growth loops, positioning against incumbents, indie-vs-giants, landing-page and
+website conversion, SaaS strategy. Monetization mechanics (paywall layout, coin economies, trial
+pricing) live in the monetization track; only the packaging levers that read as growth are here.
+
+---
+
+## Core principles
+
+### 1. Distribution is a design problem in disguise
+Don't ask users to come to you; design your product *into* the surface they already have open.
+- Raycast shipped launch-day integrations with GitHub, Jira and Slack that did **real jobs** (open your
+  assigned issues, update a ticket) rather than logo-level "we integrate with X." Every install happened
+  inside a tool the installer's teammates also use, so value was demonstrated in a shared workspace
+  with **no referral mechanics and no share-to-unlock**. `<40` people, hundreds of thousands of users,
+  $100M+ valuation [28].
+- The operative test: *where are my users already working, and how can the design meet them there?*
+  Concrete surfaces named: Chrome extension, Slack command, Notion embed [28].
+- Granola's value lands inside a meeting the user had already scheduled — no new habit to install [05].
+- Same logic on the timeline axis: Spotify's Discover Weekly arrives every Monday inside an ordinary
+  playlist, so the delivery cadence itself becomes the habit [17].
+
+### 2. Concede the commoditized layer; compete on the layer the incumbent structurally cannot ship
+When AI makes feature parity nearly free (worldwide app releases +60% YoY in Q1 2026, +80% on iOS alone;
+April 2026 +104% / +89%, per Appfigures on screen), shipping features fast is table stakes, not
+differentiation [04].
+- **Carrot Weather** (2013, solo dev, paid) beat free pre-installed Apple Weather on the same forecast
+  data by making a sarcastic character's reaction the headline element and pushing the charts below the
+  fold. 2021 Apple Design Award, a decade as a solo indie business [04].
+- **Day One** beat two free pre-installed Apple defaults (Notes, Journal) by conceding the text layer and
+  winning the envelope: auto-captured weather/location, map and calendar navigation modes, search.
+  4.8★ / 116k ratings, 2014 ADA, still shipping monthly 12 years later [04].
+- **Raycast** didn't try to out-feature Alfred (the incumbent launcher, winning on breadth and
+  customization). It competed on **clarity**: fewer options, better defaults, a bigger search bar front
+  and center [28].
+- **Spotify** didn't beat Apple/Amazon/YouTube Music on catalog or specs; it beat them on the experience
+  wrapped around the tech. "In crowded markets users switch on feelings, not features" [17].
+- **Tinder** beat Match.com's ~20-year head start, eHarmony and Plenty of Fish — all of whom had better
+  algorithms, more features and bigger budgets — by re-engineering the emotional loop [21].
+
+### 3. Niche depth means changing the interface's *shape*, not its theme
+Most top-charting apps are well-marketed clones competing on the same interface shape (200+ reviewed) [02].
+The outliers invert the category's unspoken structural assumption:
+- Tiimo: "a task is a text row" → colored illustrated blocks on a vertical timeline, sized by duration,
+  icons instead of strings. Notion/Reminders/Todoist/Microsoft To Do are literally the same object [02].
+- Cal AI: "logging food means searching a database" → the camera *is* the home page; the AI writes the
+  entry and correction is the exception, not the entry path [02].
+- MacroFactor: "shame motivates compliance" → an algorithm that *expects* you to miss, back-calculating
+  metabolism from a ~3-week trend weight. No streaks, no red numbers [02].
+- Ladder: "motivation is solo" → the whole cohort does the same workout the same day [02].
+- **The strategic point:** the mass-market leader is structurally unable to follow. Notion cannot ship
+  illustrated ADHD time blocks; MyFitnessPal cannot delete its own shame mechanics without breaking its
+  broad audience [02]. Niche-washing (leader's UI + new font, "for [niche]") wastes exactly this
+  advantage.
+
+### 4. Pick your moat deliberately — there are five on offer, and they are not interchangeable
+| Moat | Mechanism | Exemplar |
+|---|---|---|
+| **Identity / community** | Community is the home feed, not a tab; switching costs are identity costs | Strava, Whoop [11] |
+| **Accumulating per-user intelligence** | Every session trains a model of *this* user that can't be exported | Midjourney, Oura, Ramp [13] |
+| **Craft / enrichment** | Years of compounding small decisions around a commodity core | Day One, Carrot [04] |
+| **Data quality** | A UI that doesn't punish honesty produces logs competitors structurally can't match | MacroFactor [02] |
+| **Consistency as switching cost** | Behaviors become automatic; leaving means relearning | Spotify [17] |
+
+Two diagnostics that operationalize this:
+- **The exit question** [13]: *if a user left today for a competitor, what would they have to rebuild from
+  scratch?* "Nothing much" = no intelligence layer.
+- **The navigation test** [11]: *where does the social layer sit in your navigation?* Buried in a tab or
+  absent = exposed to agent displacement.
+
+### 5. Relocate from utility to identity **before** agents commoditize the task layer
+Opening diagnostic: *"If an AI could do everything my app does through a single text message, why would
+anyone open my app ever again?"* [11]. Gartner (on screen, Sept 2 2025): 40% of enterprise apps to embed
+task-specific AI agents by 2026. The reframe: **agents don't replace apps, they replace the reason users
+came to your app** [11]. Segment your base by *why they're here* — the utility cohort is gone, the
+identity cohort can't be poached — and size the utility cohort now. The timing asymmetry is the whole
+argument: founders who act now build moats that are near-impossible to crack; founders who act in a year
+rebuild from scratch against products already carved into users' identity [11].
+
+### 6. Growth loops beat growth hacks; features that unlock other features beat features that sit there
+- Raycast's loop: search bar → extension widens what the bar can do → store gives reasons to keep
+  exploring → public API keeps the store growing → back to the bar. Onboarding never ends; activation
+  continues through "wait, I didn't know I could do that too" [28].
+- Design question for every proposed feature: *does this unlock something else that already lives in the
+  product, or does it just sit there?* Standalone = utility. Unlocking = compounding value [28].
+- Ramp's version: automatic deposits (every transaction trains the AI, no conscious user effort) plus
+  integration surface (accounting, HR, Slack) — each external dependency is added switching cost [13].
+- Gabe explicitly ranks the accumulation layer above tactics: "This is more powerful than any growth hack
+  you can implement" [13].
+
+### 7. The share artifact *is* the growth engine — design it; don't hope for screenshots
+- **Find the flex** [01]: the one true thing your product already knows about the user that they'd be
+  proud to broadcast. Chapters 1–2 (Mirror, Label) can fake the output; the Flex cannot — the leverage is
+  entirely in the packaging.
+- **Fuse share and invite** [01]: Airbuds gates the fuller weekly recap behind inviting **at least 1
+  friend**, so the selfish act of showing off *is* the invite. Threshold deliberately low; reward framed
+  as permanent ("FOREVER") and recurring ("every Sunday"); the withheld archive is visible behind the gate.
+- **Ship a pre-rendered artifact sized for the destination**, not a screenshot: Duolingo's square orange
+  streak card above a WhatsApp/Messages/Instagram/X share tray [10].
+- **Design it to travel outside your app**: Wordle's grid is plain characters that paste into any message
+  thread — the cheapest possible flex loop, and the strongest evidence that the artifact matters more than
+  the surface [01].
+- **Celebrate who they are, not what they did** [17]: replace "You completed 25 tasks" with "You're a
+  hardworking night owl who does their best work after 9pm." When someone shares it they're promoting
+  themselves; your brand rides along as the vehicle. Test: does the output read "this is so me" or "this
+  is so generic"?
+- Netflix's version: engineer explicit **screenshot moments** and treat "would someone post this?" as a
+  design criterion; "97% Match" (personal, specific) beats a star rating (crowd average) [16]. Tinder's
+  full-screen match celebration became the most screenshotted app screens of the 2010s — free viral
+  marketing per post [21].
+
+### 8. Sequencing: attachment first, virality second
+"You can't engineer viral sharing if users don't stick around long enough to develop that emotional
+connection" [17]. Ladder makes the same split explicit: the TikTok → quiz funnel *acquires*; the cohort
+calendar *retains* — be clear which mechanic is doing which job [02].
+
+### 9. Fix the leaking bucket before buying traffic
+Ad spend can be ROI-positive while enormous revenue leaks out of the product [15].
+- ~20–30% of new users are lost immediately to a bad first impression (impressions form in ~50ms, trust
+  decisions in under 3 seconds, mostly visual) [15].
+- 77% of users are lost within 3 days, and the culprit is onboarding, not features or marketing [22]
+  (the same video's on-screen agenda card says 70% — the figure is inconsistent and unsourced).
+- Average SaaS site converts <4% of visitors; best-in-class ~12% — roughly triple the return from the
+  same traffic [27].
+- Corollary from the paywall data: 57.7% of new apps never cross $1,000 in revenue, and "for most of them
+  this isn't a monetization problem" — sub-$1,000 apps have a distribution or product problem, not a
+  paywall one [06].
+
+### 10. Sell the outcome, never the spec
+- Jobs held up the iPod and said "1,000 songs in your pocket" instead of 1GB / 128MB [27].
+- Rewrite rule: "real-time notifications with Slack & Jira integrations" → "close tickets 3x faster
+  without switching tabs" (his illustrative copy, not a measured result) [27].
+- Dia's marketing site uses **outcome verbs as navigation** — Write · Learn · Plan · Shop · Privacy — and
+  a headline that promises a cognitive result ("**Understand** both sides of an argument in one place"),
+  not a browser feature [20].
+- Intercom aligned the *business model* with the outcome: Fin is priced **per resolved ticket**, not per
+  seat or per conversation [20].
+- Benefit copy carries hierarchy inside a single sentence: bold the benefit clause, dim the supporting
+  clause [15].
+
+### 11. Nobody reads your page — they scan it for a reason to care or a reason to leave
+Goldilocks copy (borrowed from James Clear's Goldilocks Rule): too little copy leaves objections
+unanswered and the visitor goes shopping; too much causes analysis paralysis. It is explicitly **not**
+"write less" — that's the common misconception. It's "write what matters and nothing more," anticipating
+what the audience is thinking and pre-answering objections [27].
+
+### 12. Word of mouth is a real channel when the product accumulates something
+Midjourney: **$500M revenue in 2025, $0 VC funding, ~150 employees (>$3M revenue/employee), no traditional
+marketing budget**, started as a Discord bot with no website, no native app and no polished onboarding —
+grown purely by word of mouth, beating better-funded Adobe Firefly and DALL·E. "Midjourney didn't
+out-design Adobe; it out-accumulated it" [13].
+
+### 13. Usability is now table stakes; the differentiator moved up a layer
+In the "AI slop era" competitors ship at unprecedented scale and they all stop at usability, so emotional
+touchpoints are what's left [12]. Same claim from a different angle: as feature sets and UI converge, the
+*reception moment* (how a result is delivered) is one of the last places a product can differentiate —
+ceremony as competitive surface area, not polish, because it's craft rather than feature and therefore
+copied last [10].
+
+---
+
+## Playbooks
+
+### A. Landing-page conversion rebuild (the sequence matters: reduce first, then selectively re-add)
+1. **Collapse to one primary action.** Delete competing buttons — "watch a demo," "subscribe," "learn
+   more," "download white paper" are the usual offenders. Unbounce (18,000+ landing pages): 1 CTA →
+   **13.5%**, 2 CTAs → **11.9%**, 3 CTAs → **10.5%** [27].
+2. **Rewrite every feature line as an outcome line.** Never lead with a spec number [27].
+3. **Apply the 5-part section anatomy** to each section [27]:
+   headline (names a pain point or outcome) → subhead that adds a *new* angle, not a restatement →
+   product visual for context/contrast → proof points → clear CTA.
+4. **Run the three-question scan test** on every section, assuming a fast scroll: will they catch (a) what
+   it is, (b) why it helps, (c) what to do next? If not, the fix is delivery, not volume [27].
+5. **Put proof immediately adjacent to the CTA**, where hesitation happens — Kraftful stacks customer
+   logos (Netflix, Dropbox, Meta, Google), five stars and "50,000+ product teams analyze feedback"
+   directly under the button [27].
+6. **Then re-add intent-tiered choices** for mixed-readiness traffic (newsletter, social, help docs):
+   a visually loud primary CTA, a deliberately quieter secondary ("watch a demo," "continue with Google"),
+   and **escape valves** — chat widget, support link, save-for-later — so unready visitors exit the page
+   without exiting the relationship [27]. Tiering is expressed in *contrast*, not in count.
+7. Add a risk-reversal microcopy line under the CTA ("No credit card required") and a rating strip
+   ("4.7 / 5.0, 1,000+ reviews") [27].
+
+### B. First-impression / top-of-funnel fix (websites **and** App Store screenshots)
+Apply the same five levers to both — treat store screenshots as part of the app design, not a marketing
+afterthought [15]:
+1. Clean hierarchy · 2. Generous spacing · 3. Professional imagery that actually supports the headline ·
+4. Balanced typography · 5. Outcome-focused copy.
+Diagnostic: *"Is my top of funnel a trust builder or a trust killer?"* Give a stranger 3 seconds [15].
+Structure store screenshots in descending order of persuasion: offer → support → security [15].
+
+### C. Positioning against an incumbent
+1. Name the **commoditized layer** of your category (the data, the forecast, the text, the catalog) and
+   concede it out loud [04].
+2. Run the **category-assumption audit** [02]: (a) write down the one structural assumption every app in
+   your category shares — the shape of the primary object (row, card, feed, timer); (b) ask whether it's
+   actually true; (c) build for who the user actually is, not who you wish they were.
+3. Count the steps in the competitor's primary action, then design for one (MyFitnessPal: add food →
+   search → scroll → select → confirm = 4–5 steps; Cal AI: point and shoot) [02].
+4. Run the **Moleskine test** [04]: an IKEA spiral notebook and a Moleskine hold the same text; the
+   Moleskine wins on the back pocket, elastic band, ribbon bookmark and paper weight. Name your elastic
+   band, ribbon and back pocket. If you can't, you have no craft moat.
+5. Decide **not to out-feature** — pick clarity, defaults and one primary interaction pattern, and accept
+   that sharp opinions repel some users. That repulsion is the mechanism that manufactures believers;
+   "nobody raves about the app with the most options" [28].
+6. Write **one core principle** (Raycast's: speed / lower cognitive load / fits keyboard flow) and kill
+   anything that fails all three tests. Then ignore general usage metrics, because general usage was never
+   the goal [28].
+7. Do competitor research by **downloading and actually using** direct *and indirect* competitors' apps,
+   not reading about them; keep a references board of their product screens *and marketing sites* [12].
+8. Optional framing device: state the positioning as an analogy before naming the product — "Lovable does
+   to full-stack coding what Framer did to pixel-perfect design" as a full-screen card [29].
+
+### D. Build a flex loop (the acquisition-side growth loop)
+1. Find the one true thing your product knows that the user would brag about [01].
+2. Package it as a clean, flattering, self-contained card so no editing effort stands between feeling and
+   posting [01][17].
+3. Fuse the share action and the invite action — gate the *better* version behind inviting friends [01].
+4. Set the invite threshold at a floor that reads as trivial ("at least 1 friend") [01].
+5. Pick a recurring cadence: weekly (Airbuds, Sundays) compounds faster than annual (Wrapped) [01].
+6. Make the artifact travel natively outside your app (Wordle's pasteable grid) [01].
+7. Auto-place any status/label where it's visible without user effort — Gumroad's "Top Creator" badge goes
+   live on the profile *and* every product page, with copy framed as retroactive praise ("No action
+   needed. You've already done the hard part") and justified by *buyer* benefit so displaying it doesn't
+   read as bragging [01].
+8. Do **not** bolt a referral program onto a product with no flex: "a referral program bolted onto
+   something nobody wants to show off does zero. Nada" [01].
+
+### E. Agent-era / AI-disruption audit
+1. Run the one-text-message diagnostic on your product [11].
+2. Run the navigation test — locate the social layer [11].
+3. Run the exit question — list what a churning user would have to rebuild [13].
+4. Plot each **individual flow** (not just the whole product) on the AI Disruption Matrix [20]:
+   axis 1 utility vs experience; axis 2 repetitive/rule-based vs nuanced/judgement-heavy.
+   - *Utility + repetitive* = danger zone → rebuild-from-scratch candidate, not a polish candidate
+     (e.g. Premiere Pro timeline labour: trimming gaps, syncing, removing silences).
+   - *Experience + nuanced* = low risk (games) → invest **harder** in delight, not automation.
+   - Replace "can AI replace this interface?" with "**is the interface a means to an end, or is it the
+     experience itself?**"
+5. Note the two edge cases where invisible UI backfires: **social/privacy context** (reading a bank
+   balance aloud in a co-working space) and **information density** (investment dashboards needing
+   side-by-side scanning). Keep the UI, layer outcome automation on top — anomaly flags, trend summaries,
+   recommended actions. These are exceptions; the default assumption should be that you aren't cutting
+   enough steps [20].
+6. Inventory the data you already collect and ask which streams could be *learned from* rather than merely
+   stored; make the accumulated investment visible with a concrete counter (Oura's "1,095 nights tracked")
+   so it reads as partnership, not capture [13].
+
+### F. The 4-step product growth funnel (top → bottom, cheap → expert) [15]
+1. **The 3-second test** — design for immediate trust (see playbook B).
+2. **The friction sweet spot** — not zero friction; the right level of disclosure across the whole app.
+3. **Win mapping** — map every flow, mark the biggest wins, amplify them with emotion, then attach your
+   asks (upgrade, rate, share) to those amplified peaks. Rests on the peak-end rule: people judge an
+   experience by its most intense moment and its ending, so *when* you ask matters more than the ask.
+4. **Emotional design** — motion, feedback, human touches, so a Black Friday discount from a competitor
+   isn't enough to pull the user away.
+
+### G. Packaging experiments that move conversion without touching product or price
+Run these before rebuilding the offer — all validated A/B results (screenshots sourced to `abtest.design`;
+Gabe did not run them, and no sample sizes or significance are given) [23]:
+| Lever | Bias | Case | Result |
+|---|---|---|---|
+| Trial rendered as a dated day-by-day itinerary | Progressive disclosure | Blinkist | **+23% trial signups** |
+| Rename "free trial" → "30-day guest pass" | Framing | Headspace | **+7% new signups** from one word |
+| Free trial only on the annual plan | Decoy / asymmetric dominance | Moonly | **+39% CR, +47% revenue per 100 installs** |
+| Collapse trial activation to one tap | Cognitive load | Slopes | **+25% trial start rate** |
+| 7-day trial monthly vs 14-day annual | Anchoring | Headspace | "double-digit" lift |
+| PRO labels on already-locked content (zero new features) | Loss aversion | Mobbin | **+35% free-to-paid** |
+| Lock the *second* unit of usage (2nd language) | Scarcity at proven intent | Busuu | **+83% among engaged users** |
+| Single estimated price instead of a range | Certainty | Uber | "double-digit" lift in rides/user |
+
+Priority queue for what to test, ranked by measured LTV uplift (Adapty, on screen) [06]:
+locale/translation **62.3%** > trial structure **59.6%** > plan duration **58.7%** > plan count **57.1%** >
+price **45.5%** > visual/text **34.6%**. Structure beats price; visual polish is last.
+
+### H. Where to spend design/marketing labour in the AI era [25]
+Sort every task into three buckets and staff accordingly:
+- **Marketing creative** (social, ads): AI is "basically good to go by itself" — build a template +
+  generation system (Midjourney/ChatGPT + Figma) and hand it to the marketing team so they ship without
+  design review. Gabe did exactly this as the only designer at a major crypto bridge protocol.
+- **Websites / landing pages**: AI produces the base (Relume → Figma → Figma Sites); reserve human time
+  for the three things AI does badly — marrying copy to assets, micro-interactions, animation.
+- **Product UI**: human judgement is "truly untouched"; a generic training set produces a generic
+  solution, and your users are specific people.
+
+---
+
+## Case bank
+
+**Raycast** — $100M+ valuation, <40 people (its own copy: "35 people, 31 pets, 15 countries"), hundreds of
+thousands of users, millions raised. Four levers: integrations-as-distribution (GitHub/Jira/Slack doing
+real jobs at launch), opinionated keyboard-first UX (open, type, act; no layout customization), a loop
+(search bar ↔ extension store ↔ public API), and obsessive over-delivery for one user (Mac developers who
+live on the keyboard). Early Raycast shipped with **no formal onboarding at all**; extensions and an API
+came before a marketplace — "utility for builders over browsability for browsers." Beat Alfred on clarity,
+not features. Store install counts on screen: Kill Process 262,057; Spotify Player 183,221; Color Picker
+179,182; Chrome 157,186; VS Code 152,496; Linear 93,533 [28].
+
+**Granola** — 2-screen onboarding (work-account sign-in, microphone permission), then you close the app and
+the value arrives in a meeting you'd already scheduled. $125M Series C (March 2026) at **$1.5B, up from
+$250M — 6x in 12 months**. Note the press clipping attributes its popularity to a *positioning* advantage
+Gabe doesn't say aloud: users don't mind an app on your computer transcribing, but they do mind visible
+bots in meetings — the permission screen's copy pre-empts it: "No creepy bots join your meeting" [05].
+
+**Cal AI** — 20+ screen quiz before you ever see the product; **$35M revenue in year one**. Founder Zach
+Yadegari sold it at **$40M ARR at age 18** (earlier: a gaming site for $100k at 16); his next company hit
+$300K MRR. Two of the six early quiz questions are market research for the company disguised as
+onboarding — "where did you hear about us" (free attribution data) and "tried other apps?" (segments
+switchers vs new) [02][05][06]. His three consumer rules, on screen and never narrated: demonstrate value
+in **3 seconds or less** in any ad creative; write for a **3rd grader**; make buttons so obvious you can't
+get lost — comprehension speed as the shared driver of both virality and conversion [02].
+
+**Noom** — up to **113 onboarding screens, 10–15 minutes**, $540M raised at $3.7B. Mechanism (from the
+on-screen source, more precise than the narration): progressive commitment-building — sensitive questions
+framed with context first, expectations set and repeated, and the **paywall placed only after significant
+time and emotional investment** [05].
+
+**Midjourney** — $500M revenue (2025), **$0 VC**, ~150 employees, >$3M revenue/employee, no marketing
+budget, Discord-bot origin. The moat is a trained personalization profile of your visual taste (warm vs
+cool, maximalist vs minimal, photorealism vs painterly) that lives on their servers — moving to DALL·E
+means "starting with a stranger's eyes." Personalization is **on by default** in v7; **70%+ of active
+users adopted it within 3 months**. Supporting research: the IKEA effect (Norton, Mochon & Ariely) —
+people assign **63% more value** to things they helped create [13].
+
+**Oura** — on track for $1B revenue, a **$97M US DoD contract**, retention in the **high 80s at 12 months
+vs low 30s for other wearables** — behind a **$5.99/month subscription charged on data your own body
+generates**. Three years of wear = **1,095 nights**; the EU Data Act frees the CSV container but not the
+model trained on you: "the regulation freed the container, but it couldn't free the intelligence." Oura
+surfaces the night count so users watch their own investment grow [13].
+
+**Ramp vs Brex** — Ramp: 50,000+ companies (Shopify, Figma, Notion), millions of autonomous spend decisions
+in a single month, financial logic encoded as accumulated configuration plus integrations into accounting,
+HR and Slack. Brex, comparable design and engagement, was **acquired by Capital One in January 2026 for
+less than half its peak valuation**. "Great design got users in the door for both, but only Ramp made the
+exit door heavier." (Gabe's own note flags this as illustrative — Brex's decline has other widely-cited
+causes) [13].
+
+**Strava** — ~**$500M ARR**, **$2.2B valuation** (2025); on-screen article adds **>150M registered users**
+and **>50% new-user growth in 2024**, in a category where funding collapsed from $6.27B (2021) to $1.26B
+(2024). The home feed is **other people's activity, not your own dashboard** — an architectural decision,
+not a feature. Runkeeper and a dozen trackers had competitive or superior GPS and lost. The hiring signal:
+April 2024, CPO **Matt Salazar** from Epic Games (Fortnite/Rocket League/Epic Store growth, previously ~5
+years at Nike Run Club/Training Club) and CTO **Rob Terrell** from Zynga (~12 years, FarmVille) — gaming
+operators, not GPS engineers [11]. Separately: 180M users across 185+ countries, **14 billion kudos in
+2025**, clubs **+59% in 2024**, and a flagship stat that is the inverse of engagement maximization —
+**one hour of real-world activity per two minutes spent in the app** [07].
+
+**Strava segments** — the fix for dead leaderboards: thousands of hyper-local micro-competitions scoped by
+geography *and* age/gender cohort, so placing is plausible. Winnability is the strongest predictor of
+competitive motivation. "Engineer the size of the competition rather than inflating empty metrics" [07].
+Cohort size is a tunable growth lever, not a fixed product decision; the only quantitative anchor given is
+roughly **5–12 visible rivals** — small enough to name the person above you [03].
+
+**Whoop** — >**$800M raised**, valuation "well north of $3B" (on-screen article: **$3.6B**), **>50% of
+members still using the product daily 18 months after purchase**. Whoop Age: one number from **9 metrics**,
+plus Pace of Aging refreshed **weekly** on a **-1 to 3** scale from a trailing **30 days**. Every biometric
+is translated into identity currency, not clinical units (VO2 max 58 → **-5.3 years**). Gabe ignores his
+daily recovery score entirely but changed his sleep and step behaviour because of the weekly number — and
+became "obsessed with competing" with his wife's score. Lesson: your most-instrumented metric is probably
+not your emotionally load-bearing one [11].
+
+**Robinhood Social** (unproven bet, announced Hood Summit 2025, invite-only) — verified live trades you can
+follow and trade from. Positioning is deliberately **"I am part of an informed trading community that makes
+me a better investor,"** not copy-trading; the latter builds a leaderboard of luck and churns. Verification
+is non-negotiable when money is at stake; "Invites coming soon" is stated as a first-class page element,
+not fine print [11].
+
+**Ladder** — **500% growth in 2023** (per the RevenueCat Sub Club episode with CEO Greg Stewart, which
+credits underserved niche + product iteration + TikTok marketing, not the cohort calendar alone). Pins
+every team member to the same workout the same day; the Ladder Wall shows **4,944 completions**; a live
+"TEAMMATES WORKING OUT" avatar row with one-tap cheers sits on the workout screen itself. **TikTok → quiz
+funnel acquires; the cohort calendar retains.** 2025 iPhone App of the Year finalist [02].
+
+**Tiimo** — Copenhagen visual planner for neurodivergent/ADHD users; illustrated blocks on a timeline sized
+by duration, an icon library of thousands. Apple App Store Awards recognition in 2025 (narration says
+"won"; the on-screen Apple text lists it among *finalists* alongside BandLab and Ladder) [02].
+
+**MacroFactor** — **$72/year, 500,000 users, 4.8★, bootstrapped, five co-equal owners**, premium-only by
+design ("We will **never** serve ads"). No streaks, no red numbers, no guilt pushes. The moat claim: an
+app that doesn't punish honesty produces cleaner logging data than competitors that compete on adherence
+pressure — MyFitnessPal's user-submitted database is polluted (the on-screen evidence is one Reddit post
+showing a green-verified "Zero Sugar Cola" entry with serving size "1 slice," 240 cal) [02].
+
+**Carrot Weather / Forest / Day One** — the three-pattern indie playbook [04]. Carrot: the character's
+comedic reaction *is* the forecast; the mascot gets its own tab. Forest: put the **stake** on screen, not
+the metric — one slowly breathing tree, timer secondary, "Give up" demoted to a low-contrast link; **2M+
+real trees planted**, 2,500 virtual coins → one real tree via Trees for the Future. Day One: four
+navigation modes (list, calendar, map, media) over one commodity content type; 4.8★ / 116k ratings.
+
+**Spotify Discover Weekly** — launched July 2015; 30 unheard tracks every Monday delivered as an ordinary
+playlist so the recommendation stack never surfaces as UI. **100+ billion tracks streamed**, **56M+ new
+artist discoveries weekly**; subsequently copied by Apple Music, YouTube Music and Amazon Music. The
+counter-example is early **Pandora**, which made users train the recommender with thumbs up/down — "doing
+homework instead of discovering music" [17].
+
+**Spotify Wrapped** — the data exists year-round; Spotify chooses to gate it, release it annually, and wrap
+it in sequential slides ending in a share prompt. On-screen third-party figures: **~60M stories shared
+(2021)**, **156M+ users engaged the following year**, **+461% tweet volume 2020→2021** [17]; 2025 edition
+reached **200M users in the first 24 hours, +19% YoY** (the prior year took 62 hours) [10]. The warning:
+teams copy Wrapped by building a data summary and miss that **"Wrapped is a ceremony, not a report"** — one
+idea per slide, huge type contrast, editorial voice, no charts [10].
+
+**Netflix** — ~$400B. Reclassified infinite choice as a **psychology problem, not a technology problem**.
+Skip Intro: usage data showed **15% of viewers manually fast-forwarding**, so they built AI intro detection
+(pre-hype) and shipped one button — **136 million presses/day (2022)**, **195 years of cumulative user time
+saved per day**. Personalization engine (2015) serving **230M users** via **1,300+ micro-genres**, with the
+*reasoning* exposed inline ("Your taste preferences created this row"); **80% of viewing comes from
+recommendations**. Viral UX: "Are you still watching?", the ta-dum, "97% Match" as a dating-trophy number,
+category names in human language ("Because you watched…") — "Netflix and chill didn't happen by accident"
+[16].
+
+**Tinder** — 75M MAU, $2B+ annual revenue, 1.6B swipes/day, 26M matches/day, **11-minute sessions vs ~3 for
+competitors**, 9.6M paying subscribers. Sub-60-second onboarding via Facebook login + auto-imported photos;
+double-blind matching so rejection is invisible ("if you don't know you're being rejected, it doesn't hurt.
+If it doesn't hurt, you keep playing"); full-screen match celebration as free viral marketing; monetization
+deferred until engagement and then triggered at the moment the matching desire appears — Boost after
+matches start, Gold when curiosity about likers peaks. Paywall copy quantifies the benefit ("You're 3x more
+likely to get a match") [21].
+
+**ReelShort** — **35.7 min/day per US user vs Netflix 24.8, Prime Video 26.9, Disney+ 23** (Omdia/Sensor
+Tower). Growth-relevant mechanics: the **graduated gate** (episodes 1–5 free with no account at all → 6–8
+gated by a 30-second ad → 9+ gated by coins at the cliffhanger; "attention, effort and then money"), and
+**exit-ramp counting** — Netflix's mobile flow has at least six decision points between opening the app and
+playback; ReelShort's is open-app-and-it-plays. "The fewer exit ramps you build, the more momentum your
+product carries." Caveat Gabe doesn't surface: ~**1.2M US MAU vs Netflix's ~12M** — the win is
+minutes-per-user, not audience or revenue. Global microdrama revenue $11B (2025) → $14B projected (2026);
+microdrama downloads >2.3B in 2025 (>2x prior year) while traditional streaming downloads fell >4% [08].
+
+**Intercom / Fin** — category leader with PMF, brand equity and thousands of customers; per CPO Paul Adams
+they **paused the roadmap within a week of ChatGPT launching** and built an AI-native product aimed at one
+outcome — instant resolution. Priced **per resolved ticket**. Now Intercom's fastest-growing product ever
+and the company's main focus [20].
+
+**The Browser Company (Arc → Dia)** — killed a beloved, innovative, popular product publicly (and took the
+backlash) rather than stapling AI onto an old interaction model, then rebuilt around answers instead of
+navigation. The BlackBerry test: BlackBerry owned ~half the US smartphone market and kept refining
+keyboards and scroll balls for a world users had already left [20].
+
+**Kraftful** — the worked landing-page example: outcome headline ("Get insights to build products people
+love"), subhead adding a *new* angle plus an objection-handler aimed at AI skeptics ("analyzed in minutes,
+no hallucinations"), inline email field attached to the only saturated-colour button, a deliberately
+quieter "Continue with Google," and proof directly beneath — Netflix/Dropbox/Meta/Google logos, five stars,
+**"50,000+ product teams analyze feedback."** Nav items (Pricing, Team, Blog, Case Studies) live in the
+header rather than competing in the hero [27].
+
+**Unbounce + the Columbia jam study** — 18,000+ landing pages: 13.5% / 11.9% / 10.5% for 1 / 2 / 3 CTAs.
+Psychological engine: 24 jam varieties → **3% purchased**; 6 varieties → **30%** — a ~10x lift from
+removing options. (Gabe narrates it as a "candy shop"; the on-screen paper is the Iyengar & Lepper
+supermarket jam study. Neither dataset is dated, linked, or controlled for industry or traffic source) [27].
+
+**Airbuds** — a widget showing what friends are listening to, with a weekly Sunday recap. Real data,
+flattering packaging, and the unlock gate fuses share with invite: "Invite at least **1 friend** to get
+your Weekly Recap every Sunday," with the user's own withheld archive visible and dimmed behind the gate.
+Investor Alexis Ohanian: music "has never just been about access. It's identity" [01].
+
+**16 Personalities** — **1 billion+ tests taken**; a ~10-minute quiz returns a four-letter type plus a
+flattering archetype (INFJ "The Advocate"). Every type gets a character illustration, a group identity
+("Diplomats") and a positive blurb — **there is no losing type**. Users put it in dating bios and use it as
+self-explanation; churning would mean surrendering a piece of self-concept. "People don't churn out of
+their own identity" [01].
+
+**Starcrossed** — solo founder, distributed off her own TikTok. Heavy extraction (birthday, **exact** birth
+time, birth city) → theatrical loader ("reading your chart") → paywall at peak sunk cost → soulmate
+sketch. The app's own FAQ admits each planet has **two** pre-selected zodiac options — the result was
+assembled, not discovered — and the locked screen literally shows two empty cells per planet. It also runs
+a flex loop Gabe never mentions: "Every time you Cross with a friend, a piece of your soulmate's identity
+is unlocked." The reveal screen is designed as UGC bait; the ad creative is the reveal artifact and nothing
+else [01].
+
+**Badge retreats (LinkedIn 2024, Foursquare 2014, Google News)** — three platforms removed points/badges/
+leaderboards. LinkedIn's own statement blames **automatic** awarding for making quality standards
+unmaintainable (people gamed it by commenting on AI-generated posts); Foursquare's data showed gamification
+drove **check-ins but not the discovery behaviour the business needed** — the mechanic inflated the vanity
+action, not the revenue action; Google News offered **500+ badge types**. Yu-kai Chou: PBL is "the
+scoreboard of a game, not the game itself" [07].
+
+**Peloton** — members who engage socially work out **15% more frequently** (note: the cited source credits
+community; Gabe re-attributes the mechanism to competence feedback — that reinterpretation is his) [07].
+
+---
+
+## Anti-patterns
+
+- **Buying traffic before fixing the bucket.** 20–30% leak at first impression means even ROI-positive
+  spend loses most of the available revenue [15].
+- **Stacking multiple equal-weight CTAs.** Each measurably costs conversion; extra options create
+  hesitation, not empowerment [27].
+- **Spec-led messaging and feature lists.** The visitor can't map them to their own life [27].
+- **"Just write less."** Cutting past the point where objections are answered makes people hesitate and go
+  look at an alternative [27].
+- **Applying the one-CTA rule dogmatically.** For complex SaaS with mixed-readiness traffic, removing the
+  demo / learn-more path can hurt [27].
+- **Trying to change users' habits instead of joining them.** Adoption gets expensive when you pull people
+  out of tools they already have open [28].
+- **Referral tricks and share-to-unlock as a growth substitute** for actual usefulness in the right place
+  [28] — and **referral programs bolted onto a product with no flex**, which "do zero. Nada" [01].
+- **Out-featuring an incumbent.** They are best-resourced to win that game; compete on clarity and defaults
+  [28][02].
+- **Niche-washing a mass-market interface** — the leader's UI with a new font, sold "for [niche]". The
+  niche's whole advantage is that the leader *can't* restructure; copying their structure wastes it [02].
+- **Bolting new tech onto the old UI.** Adding a camera button to a search-first calorie tracker is a
+  feature the incumbent copies in a sprint; it resets nothing [02].
+- **Padding AI features onto a legacy interface** — a facelift inside a paradigm that's being replaced [20].
+- **Protecting a working playbook because it currently works.** Intercom had PMF, brand equity and
+  thousands of customers — all reasons to defend the old model — and threw it out anyway [20].
+- **Social features as a bolt-on / buried in a tab.** "Strava is not a fitness tracker with social features
+  bolted on." If the social layer sits under the utility layer architecturally, it's decoration [11].
+- **Waiting for churn before adding community.** By then you're building from scratch against products
+  already embedded in users' identity [11].
+- **Treating agents as "just a feature/assistant"** — the wrong frame entirely [11].
+- **Optimizing engagement alone.** Engagement is a treadmill: it keeps users running but doesn't stop them
+  jumping onto a different one, and every competitor can copy your animations [13].
+- **Storing data without learning from it.** Warehouses don't produce lock-in; regulation can force the
+  container open [13].
+- **Making users invest consciously.** If the deposit requires deliberate effort, most won't make it [13].
+- **Chasing virality before attachment** [17], and **assuming viral features are accidents**, so you never
+  invest in shareability [16].
+- **Generic milestone bragging** ("You completed 10 tasks", "You're a power user") — hollow, could describe
+  anyone, reveals nothing except that the user used an app [17].
+- **Making the technology the hero.** The stronger the tech, the greater the temptation to expose it; a
+  separate "AI inbox" or AI search page adds learning cost and kills the invisibility [17].
+- **Forcing users to train your model** (early Pandora) [17].
+- **Random-timed asks** — paywalls, tutorials and rating prompts fired when users are frustrated or
+  confused rather than at peak satisfaction [15].
+- **Gamifying the wrong action.** If your mechanic inflates a vanity metric it will succeed at inflating it
+  and fail at the business (Foursquare) [07].
+- **Chasing general usage metrics when you've chosen a niche** — they won't reflect the user you designed
+  for [28].
+- **Assuming AI-speed shipping is a strategy.** With releases up ~80–104% YoY, fast feature shipping is
+  table stakes and therefore not a differentiator [04].
+- **Bolting personality onto the edges** — friendly empty states, confetti on first save, a mascot on the
+  loading spinner — while the core surface is still a chart [04].
+- **Overhauling the offer when the packaging is the problem.** Blinkist changed nothing about the trial;
+  Mobbin shipped zero new paid features [23].
+- **Treating consistency as a nice-to-have visual concern.** It's the moat, not the paint [17].
+
+---
+
+## Decision guide
+
+**IF** your product depends on heavy personalization in a deep-pain niche **THEN** a long onboarding is
+justified — it earns the personalized first run *and* filters for high-intent buyers (Cal AI: 20+ steps,
+$35M year one). Make the personalization **visible on the first post-onboarding screen**, or the user paid
+the toll and got the generic product anyway [05].
+
+**IF** your product is not personalization-dependent **THEN** the fix for bad retention is *less*
+onboarding, not more (Granola: 2 screens, 6x valuation in 12 months) [05].
+
+**IF** you are choosing between "short onboarding" and "long onboarding" **THEN** you are on the wrong axis.
+Optimize **time-to-tangible-value**; brevity is only its proxy. Few screens that deliver nothing is still a
+failure [05][22].
+
+**IF** a free, pre-installed, first-party app stores the same content you do **THEN** concede the data layer
+and compete on the envelope: auto-captured context, navigation modes over the same content, search quality,
+animation timing, metadata placement [04].
+
+**IF** your category leader is winning on breadth and customization **THEN** compete on clarity: fewer
+options, better defaults, one primary interaction pattern, and accept that sharp opinions cost you some
+users — that's the mechanism that produces believers [28].
+
+**IF** your traffic arrives from mixed sources at mixed readiness (newsletter, social, help docs) **THEN**
+keep one visually dominant primary CTA and add a *subordinate* secondary plus escape valves — not a second
+equal-weight button [27].
+
+**IF** your page has one clear high-intent audience **THEN** delete every competing CTA; each extra one
+costs ~1.5pp of conversion in the Unbounce data [27].
+
+**IF** your product's value is a resolution or an outcome rather than a session **THEN** price on the
+outcome (Fin: per resolved ticket), not per seat or per interaction [20].
+
+**IF** a flow is utility + repetitive/rule-based **THEN** treat it as a rebuild-from-scratch candidate, not
+a polish candidate. **IF** it is experience + judgement-heavy (games, emotional products) **THEN** the
+interface *is* the value — invest harder in delight [20].
+
+**IF** the flow is privacy-sensitive or information-dense **THEN** keep the UI visible and put AI in a
+background support role (anomaly flags, trend summaries, bill prediction), not a voice/invisible interface
+[20].
+
+**IF** an agent could complete your product's core task over a text message **THEN** your roadmap is
+answering the old question ("can users accomplish the task?"). Move to identity/community/ritual now,
+while it's cheap [11].
+
+**IF** you have enough per-user data density **THEN** build the intelligence layer (accumulating trained
+personalization, on by default, deposits automatic). **IF** you're pre-PMF with a couple hundred users
+**THEN** this framework doesn't apply yet — Gabe never addresses that case [13].
+
+**IF** you want a growth loop **THEN** find the flex first and fuse share with invite; **IF** you have no
+brag-worthy artifact **THEN** a referral program will do nothing — build the brag first [01].
+
+**IF** your competition is a global leaderboard nobody can win **THEN** shrink the cohort (geography, age,
+gender, friend circle, route) until winning feels plausible — roughly 5–12 visible rivals [07][03].
+
+**IF** conversion is flat **THEN** test structure before price: locale (62.3% LTV uplift) > trial structure
+(59.6%) > plan duration (58.7%) > plan count (57.1%) > price (45.5%) > visual/text (34.6%) [06].
+
+**IF** your app has never crossed ~$1,000 in revenue **THEN** you likely have a distribution or product
+problem, not a paywall problem — don't start with monetization tactics [06].
+
+**IF** the task is marketing creative **THEN** hand it to AI + templates almost entirely; **IF** it's a
+landing page **THEN** let AI build the base and reserve humans for copy/asset orchestration,
+micro-interactions and animation; **IF** it's product UI **THEN** keep human judgement in the loop [25].
+
+**IF** you're about to launch a viral/shareable feature **THEN** check retention first — sharing moments
+don't work on users who haven't stuck around long enough to care [17].
+
+---
+
+## Contradictions and tensions across the notes
+
+1. **Short onboarding vs. deliberate effort.** [22] and [05] argue for time-to-value and cutting friction;
+   [01] argues the opposite — "stop treating long onboarding as friction… ask for more, not less," because
+   effort justification amplifies the perceived value of the payoff. [05] partially reconciles this (length
+   is justified only if it buys personalization or a buyer filter), and [06] adds a third framing: the
+   binding constraint is **trust, not brevity**, which is why a 28-screen flow can outperform a 3-screen
+   one. Treat "short by default" as a prior, not a law.
+
+2. **Delete friction vs. the friction sweet spot.** [08] says count and delete exit ramps ("every
+   unnecessary tap is an exit ramp"); [15] says removing *all* friction is wrong — over-stripped onboarding
+   confuses users and they quit just like over-long onboarding does. [08]'s own caveat is that its
+   deletions are downstream of a format choice (60-second vertical episodes), so the transferable part is
+   "map every decision point," not the specific deletions.
+
+3. **Consistency as moat vs. killing your own paradigm.** [17] argues design consistency manufactures
+   switching costs and should be budgeted as moat-building spend; [20] argues you should be willing to
+   sunset a beloved product (Arc) rather than iterate inside a dying paradigm. [17]'s note flags the
+   unaddressed downside: consistency-as-lock-in also constrains your own ability to redesign.
+
+4. **What the moat actually is.** Four videos each nominate a different primary moat — identity/community
+   [11], accumulated per-user intelligence [13], craft/enrichment [04], data quality [02] — and each is
+   argued as *the* durable one. None is tested against the others. Treat them as a menu (principle 4), not
+   a ranking.
+
+5. **Engagement maximization vs. Strava's flagship metric.** Most of the corpus optimizes for time in app;
+   [07] approvingly cites Strava marketing "one hour being active for every two minutes spent on the app" —
+   the inverse.
+
+6. **Numbers that don't reconcile.** "99% of apps don't reach real commercial success" [15] vs "57.7% of
+   new apps never cross $1,000" [06]. "77% of users lost in 3 days" [22][05] appears with an on-screen
+   graphic saying 70% in the same video [05]. ReelShort "beats Netflix" on minutes/user while having ~10x
+   fewer US MAU [08]. Spotify is called a "$2 billion empire" and valued at "$25B+ at IPO" in the same
+   video [17].
+
+7. **Evidence quality.** Almost none of the causal claims are tested: Granola's growth is attributed to
+   2-screen onboarding while the on-screen press clipping credits a positioning advantage (no bots in
+   meetings) [05]; Ladder's own source credits 500% growth to niche + iteration + TikTok, not the cohort
+   calendar [02]; Brex's decline has widely-cited causes other than missing investment loops [13]; the
+   Unbounce segments aren't controlled for industry, offer type or traffic source [27]. Cite the numbers,
+   not the causation.
+
+---
+
+## Sources
+
+- `[01] The Twisted Psychology Behind Top 1% Apps — https://www.youtube.com/watch?v=nQKWYmrixCU`
+- `[02] The Weird Design Playbook of 6 App Outliers — https://www.youtube.com/watch?v=HA4yB_sd62Q`
+- `[03] Why Leaderboards Kill App Retention (How To Fix It) — https://www.youtube.com/watch?v=BxhsCu9hNpY`
+- `[04] How Top 1% Indie Apps Beats Giants — https://www.youtube.com/watch?v=MAvKHToGiVY`
+- `[05] How To Solve The App Onboarding Paradox — https://www.youtube.com/watch?v=Aa89MC8jX2c`
+- `[06] I Studied 10,000 Paywall Screens (THIS Makes People Pay) — https://www.youtube.com/watch?v=sYRhXB_ZcLI`
+- `[07] I Studied 500+ Gamified Apps (Here's What Actually Works) — https://www.youtube.com/watch?v=LXX_qOA5D8E`
+- `[08] How a Chinese Unicorn App Out-Addicts Netflix — https://www.youtube.com/watch?v=7dCZrtH15zc`
+- `[10] The 3-Stage Trick Behind Every Addictive App — https://www.youtube.com/watch?v=uoLpH_20qKw`
+- `[11] Why Most SaaS Companies Will Die in 2026 — https://www.youtube.com/watch?v=ZXWUg1e1PAs`
+- `[12] Our World Class App Design Formula — https://www.youtube.com/watch?v=wmTkiF23GRQ`
+- `[13] The New Way Apps Dominate in 2026 — https://www.youtube.com/watch?v=yYs7iv81Ppk`
+- `[15] Give Me 13 Minutes & I'll Grow Your App — https://www.youtube.com/watch?v=v1f9iYZ2GhQ`
+- `[16] How Netflix Made Us All Binge (Design Breakdown) — https://www.youtube.com/watch?v=hIm3d3JAsPw`
+- `[17] Viral Design Tricks from Spotify (Founder Playbook) — https://www.youtube.com/watch?v=Tpg0pxKHrCA`
+- `[20] The Future of App Design is Invisible — https://www.youtube.com/watch?v=zRUoPIwCxfw`
+- `[21] How Tinder's Design Hacked Society — https://www.youtube.com/watch?v=4yKtlOWClhA`
+- `[22] The Hidden App Growth Killer (How To Avoid It) — https://www.youtube.com/watch?v=55hDj88zKa8`
+- `[23] Copy These SaaS Growth Tricks (It'll Blow Up Your Business) — https://www.youtube.com/watch?v=puldme__Ckk`
+- `[25] Do Companies Really Need Designers Anymore? — https://www.youtube.com/watch?v=bOm9GDv96Tk`
+- `[27] Give Me 10 Minutes & I'll Boost Your Website Conversions — https://www.youtube.com/watch?v=TIfiy1fodDs`
+- `[28] How Raycast Reached a $100M+ Valuation (As a Small Startup) — https://www.youtube.com/watch?v=eWGNjTLK84c`
+- `[29] 5 Tools I Wish I Found Sooner (For Designers) — https://www.youtube.com/watch?v=qmJVN2158Mk`