Video summary

$22M на бутстрапе и 10M инсталлов. CEO Moonly о ферме креаторов, отказе от триалов и пейволах

Main summary

Key takeaways

Business

Bootstrapped growth + “slow growth” for category/rewrites

  • First $20M revenue before investments: positioned as able to scale without VC funding.
  • End of 2024: they intentionally slowed growth to prioritize:
    • Product readiness
    • Major replatforming / rewrites

Key actions during the slow-growth period:

  • Bought/produced top content via millions of dollars invested in content (hundreds of craftsmen).
  • Built an app earlier, but learned user needs changed—requiring a complete rewrite.
  • After raising their first seed/preset ($3M):
    • Spent the full annual budget
    • Temporarily doubled team size for a year, then reduced.

Strategic product shift: “Nextgen” (AI-rails) instead of minute-based chat loops

They explicitly avoid the monetization model common in their category:

  • Competitors monetize live chats by time (“sell minutes”).
  • They describe this as creating a business-model vs user-goal conflict:
    • encourages “hooking”
    • and slows resolution

Goal: “healthy loops”

  • Build healthy loops that increase practice/insight rather than time-wasting.

AI architecture / process (multi-agent system)

They built a multi-agent system that:

  • understands user emotional/psychological intent
  • anticipates requests
  • routes users to the right experiences (described as “competently routes you” to practices/content)

Dream Book feature

  • Users share dreams
  • The system draws interpretation and provides Jungian interpretation
  • Collects dream/context data across esoteric + psychological domains to improve personalization

Rollout status: agent is beta, with tens of thousands of dialogues per month.

Metrics & KPIs mentioned (growth + unit economics + retention/momentum)

  • 70M views/month
  • CPM < $0.30
  • Target: “Turn 70M views into a billion” (stated goal by end of the year)

Monetization / engagement indicators (examples)

They track signals like:

  • Return to chat
  • Purchases from chat
  • Conversion after dialogue (e.g., “Do you buy a subscription after communicating with her?”)
  • Paid additional feature recommendations

They also monitor sentiment/outcome by summarizing dialogues at scale:

  • human-readable summaries currently
  • automated summarization once out of beta

Retention by cohort (payments experiment)

  • Users choosing alternative payments have cohorts that live 2–3x longer (explicitly stated as ~2–3x).

Product & onboarding playbook: rework onboarding + attribution-driven scenarios

2025: “sawed up everything”

Not just surface UX—major rebuilding including:

  • Rebuilt onboarding and the user journal
  • Rewrote onboarding cost and user management
  • Built custom marketing attribution to match ad traffic users to downstream behavior

Strategy outcome

  • Enables custom scenarios and store pages by intent (e.g., meditation/tarot creative pages)

  • Onboarding text and prompts are designed to trigger relevant features, not merely describe them.

Experimentation principles

  • Avoid “talk about features first” (they tried it early; sometimes worked but wasn’t durable).

  • Use pain-point discovery + customer interviews continuously.

  • Mix features with activating knowledge / “aha” moments (example: dream recall question → then reveal Dream Book).

Pseudo-personalization

  • Character selection (persona) + reviews + before/after framing.

Character/tool selection safety

  • They test different characters (e.g., “master” selection) to avoid mismatch and repulsion risk.

Content + marketing execution: “creator farm” with actor-based production + AI templating

Core marketing model: “buy an actor”

  • Create clean actor accounts
  • Warm them up
  • Script/perform hooks (instead of relying on influencer audience dynamics)

Scale economics

  • Took 6+ months to build the creator farm
  • Now: ~3 people + tools + sitework
  • Total cost: $15–17K/month (salaries + tools + ops)
  • They cite heavy early learning costs: hundreds of thousands of dollars burned discovering what worked

Creative pipeline & automation

They use AI tooling (described as “Spider”, similar to Freepik/CoffeeUI workflows) to break videos into “bricks”, and replace:

  • face (conditional face replacement for generated outputs)
  • background setting
  • text blocks and prompts

Important packaging/distribution rule:

  • Keep interface logic correct—especially the app logo:
    • it must appear in the app UI layer
    • otherwise platform distribution can be reduced.

Repurposing a hook

  • One paid actor hook can expand into ~100 derivative videos via AI variations.

Performance claim

  • 70M views/month maintained with CPM < $0.30
  • They believe it’s inexpensive and compounding (brand + organic pickup), but:
    • they cannot isolate causality because it also affects brand/CPIs.

Attribution + creative optimization process

They argue marketing shouldn’t be a “black box”:

  • The CEO performs daily deep dives into:

    • creatives
    • messaging
    • pain points during marketing calls

  • They redesigned the workflow so engineers/CPO can build on common-sense logic rather than “classic” marketer constraints.

Attribution “lab”

  • Pulls real-time data into decisions (described as “real-time Facebook posts → decisions”).

Measurement improvement mentioned

  • Rejected probabilistic measurement (“MPs” with ~60% probabilistic quality and multi-day timing)
  • Switched to near real-time quality, and:
    • data quality doubled
    • “ACES doubled in size overnight” (as stated)

They also referenced building a content factory inspired by other companies (Plurio, founders Seva and Kirill), but decided it was too early to fully implement.

Monetization experiments: removal of trials + alternative payments design

Trials

  • They say trials “die” because users develop a habit of canceling immediately.
  • They’ve not offered trials for a long time.
  • Instead, they experiment with feature presentation and plan structuring.

Pricing/offer experiments

  • Bold animations vs simple lists
  • Plan mixing
  • Discounts and “green dog” / anchor-like structures

Alternative payments (US rollout)

Availability:

  • iOS: last summer
  • Android: spring
    • they note Google SDK constraints limited implementation; conversion was poor

UI iteration to reduce selection bias:

  • Even when alternative payments exist, they changed UI to avoid multiple-button selection bias.
  • Early attempt: adding an alternative payment button reduced conversion.
  • They found a minimal/simple design that made it profitable.

Outcome:

  • ~half of users choose alternative payment method
  • No need to push further.

Key “do/avoid” recommendations implied by their execution

Do

  • Build product loops that improve user outcomes (microhabits), not time-based hooks.
  • Invest heavily in content quality as a moat, then amplify delivery with AI.
  • Run onboarding as a scenario engine (traffic source → personalized first-session flow).
  • Treat marketing + attribution as engineering-grade systems (real-time data; no black boxes).
  • Scale creative output with repeatable automation:
    • actor hook → AI variations → interface-correct packaging

Avoid

  • Competing in “minute-based live chat” economics that conflict with user goals.
  • Over-optimizing cosmetically with risky automation:
    • they tried AI-generated character imagery
    • it caused edge-case ugliness that made some users avoid their own image
    • they cut it.
  • Relying on probabilistic attribution providers when you need real-time learning loops.

Frameworks / processes / playbooks referenced

  • Healthy loops vs dopamine traps (loop design philosophy)
  • Onboarding filter framework:
    • show features only after answering “Why do I need this?”
    • pain-point/value alignment loop
  • Customer development loop:
    • continuous interviews → extract pain points → update onboarding and feature reveals
  • Attribution-driven experimentation:
    • traffic → user mapping → scenario personalization → creative iteration
  • Creator farm operating model:
    • actor-based production + AI templating + interface-safe distribution packaging

Presenters / sources

  • Presenter / interview subject: CEO Moonly (company brand name not explicitly provided in subtitles)

  • Other named/credited people & sources within the talk:

    • Sergei Tokarev (Rush Foundation) — investor mentioned
    • Seva and Kirill — founders associated with Plurio (used as inspiration)
    • Google — referenced as providing tooling/experts for practice selection

Original video