§ 01 · Expanded Marketing Proposal · July 2026 · v1.0
From affiliate-first
to a compounding revenue system.
Candy.ai already proved the product can print cash. AYMI builds the demand infrastructure that turns a high-velocity affiliate machine into a multi-channel subscription engine with owned creative, owned audiences, and defensible LTV.
§ 02 · The opening frame
You built the category winner.
Now build the system that keeps it.
Product-market fit is not the hard part anymore. Diversified, brand-safe, LTV-defending acquisition is.
Candy.ai (EverAI Limited, Malta) launched around September 2023 as a web-first AI companion platform — chat, voice, image generation, short video, custom characters. Public reporting and partner case studies place the brand among the commercial leaders in the AI-girlfriend category, with subscription pricing ($13.99/mo monthly · $8.99/mo quarterly · $3.99/mo annual) plus token top-ups for media features, and a reputation for aggressive affiliate economics (reported up to ~40% lifetime share).
That is an extraordinary early-stage outcome. It is also a concentration risk. Growth that scales primarily through affiliates and SEO partners is real revenue — and it is rented. When a partner channel cools, policy shifts, or a clone undercuts commission, the top line moves with it. Paid social in this category is constrained (Meta/TikTok adult-adjacent policy, brand-safety walls, creative fatigue), which makes a generic “just buy more ads” brief useless.
AYMI builds the missing layer: a Candy Revenue OS — brand-safe creative systems, policy-aware paid acquisition, lifecycle that protects LTV after the free trial, conversion infrastructure that routes intent by persona, and measurement that separates trial starts from paid months and token attach. The goal is not “more traffic.” It is a compounding subscription business that can grow without betting the company on any single partner channel.
Scope note. This draft is high-level and prior to speaking with the Candy.ai / EverAI team. Traction figures cited from third-party coverage (ARR, user counts, affiliate terms) are directional context only — not verified internal metrics. We calibrate every number on the scoping call.
§ 03 · Where you are · where this takes you
The same product.
A different operating shape.
Eight axes of the growth surface — current state vs. the system AYMI stands up.
| Axis | Current (public read) | Future (AYMI build) |
|---|
| Acquisition mix |
Affiliate + SEO partner heavy; paid social secondary or constrained by category policy |
Diversified stack: brand-safe paid (Meta/TikTok/YouTube where policy allows) + owned search + affiliate as a channel, not the only channel |
| Creative system |
Product-led screenshots and character gallery; partner toolkits vary by affiliate |
Weekly creative cadence with Meta-safe / brand-safe frames, persona hooks, and a kill/scale board — not one-off assets |
| Funnel |
Strong free-trial → paid subscription path; token economy for media features |
Persona-routed landers, trial-to-paid CRO, and token-attach paths instrumented as separate conversion events |
| Lifecycle |
Product engagement is high; lifecycle marketing depth less visible from outside |
Trial nurture, dunning, win-back, annual-plan upgrade, and token-upsell flows with hold-discount discipline |
| Brand surface |
Category leader in “AI girlfriend” search; reputation mixed by design (NSFW + mainstream curiosity) |
Dual-register brand system: mainstream-safe public brand for paid/PR + product-true experience post-login |
| Measurement |
Partner-attributed signups; limited public view of blended CAC / LTV by source |
Source-level CAC, trial→paid, month-2 retention, token attach, affiliate vs paid contribution — one weekly decision dashboard |
| Policy risk |
Adult-adjacent category faces ad-account and platform volatility |
Policy playbook: creative rules, account structure, backup channels, and claims discipline so growth does not depend on one platform’s mood |
| Resilience |
Bootstrapped / revenue-funded growth is a strength; single-channel dependence is the soft underbelly |
A system that compounds: creative IP, owned audiences, lifecycle, and paid that can flex when partners or platforms shift |
§ 04 · Directional growth benchmarks
What a healthy system
should produce.
Illustrative 12-month operating targets for a scaled subscription companion brand — calibrated on the Growth System engagement shape, not contractual guarantees.
3+
Meaningful acquisition channels (paid · organic · partner)
≤35%
Share of new paid from any single partner channel
+25–40%
Trial → paid conversion lift via CRO + lifecycle
+15–25%
Month-2 paid retention via onboarding + dunning
Weekly
Creative refresh cadence (kill / scale board)
4–6
Live brand-safe concept families in paid
3:1+
Target LTV:CAC floor on paid cohorts
1 view
Weekly decision dashboard (source · creative · LTV)
Targets are directional. Final ranges lock after we see Candy.ai’s real source mix, trial→paid rates, token attach, refund/chargeback rates, and ad-account health. Third-party ARR/user claims are not used as baseline.
§ 05 · The three buyers
Different motives.
Same subscription.
Persona-led acquisition beats a single “AI girlfriend” pitch — especially under paid-platform policy.
Persona 01
The Lonely Evening
Adults seeking low-pressure companionship — chat that remembers, voice that softens the night, zero social performance. Privacy and discretion are non-negotiable.
Pain language
“I just want someone to talk to without the apps.”
Entry wedge
Brand-safe “always there” creative · trial → paid · night-time lifecycle
Persona 02
The Character Builder
Creators and role-players who treat Candy as a sandbox — custom personas, story arcs, image/video generation. Tokens and customization are the product, not a side feature.
Pain language
“Other bots feel scripted. I want control.”
Entry wedge
Create-your-own demo · feature education · token attach
Persona 03
The Curious Switcher
Users hopping from Character.AI, Replika, or clone apps — comparing realism, uncensored range, price, and privacy. They convert on proof and plan clarity, not hype.
Pain language
“I’ve tried three of these. Which one actually sticks?”
Entry wedge
Comparison landers · plan clarity · annual upgrade path
All three land in the same subscription + token economy. Creative and landers speak their language; the backend is one product system. Paid creative for Personas 01 and 03 stays deliberately brand-safe; Persona 02 can go deeper on product features post-click.
§ 06 · The most important expansion
The Candy Revenue OS.
One system that diversifies acquisition, protects LTV, and survives platform policy — without dulling the product that made Candy win.
The anchor is not a single landing page or a bigger affiliate payout. It is an operating system: brand-safe demand in public, product-true experience after login, and measurement that makes weekly capital allocation obvious.
The compounding loop
Every stage produces data that feeds the next creative and budget decision.
01
Persona hooks
Brand-safe paid + organic + partner
02
Routed landers
Trial · plan · create-your-own
03
Trial → paid
CRO + day 0–7 nurture
04
LTV defense
Onboarding · dunning · annual · tokens
05
Creative loop
Weekly kill / scale board
§ 07 · Paid acquisition
Creative cadence beats
creative budget.
- Brand-safe concept families — companionship, creativity, privacy, “always there” — never explicit claims or imagery in paid surfaces.
- Weekly refresh cadence — 4–6 live concepts, structured kill/scale, not monthly “campaign launches.”
- Account architecture — separation of testing vs. scaling, backup assets, and policy playbooks so one rejection does not freeze growth.
- Landing match — every concept maps to a persona lander (trial, create-your-own, plan compare).
- Affiliate coexistence — paid fills gaps partners cannot (brand search defense, retargeting, new geos) rather than competing blindly on the same keywords.
§ 08 · Content & SEO
Own the comparison layer
and the education layer.
- Comparison & review cluster — Candy vs. category alternatives, plan explainers, token economics — written in human language, not SEO sludge.
- Feature education — voice, image, create-your-own, privacy — content that reduces support load and lifts trial quality.
- AEO / GEO readiness — structured answers LLMs and answer engines can cite without inventing product claims.
- Selective PR — founder/operator narrative (subscription economics, web-first distribution) for business press; product narrative for tech/culture — kept cleanly separate.
§ 09 · Email & lifecycle
The retention spine.
- Day 0–7 trial nurture — activation milestones (first chat, first image, first voice), not generic “come back” spam.
- Trial → paid conversion series — plan clarity, annual value, privacy reassurance.
- Dunning & failed-payment recovery — quiet, high-ROI revenue defense.
- Win-back & pause alternatives — before hard cancel; annual save offers only after value is established.
- Token attach flows — for Character Builders without training the whole base to expect discounts.
- Hold-discount discipline — no panic offers in early lifecycle that train churn-for-coupon behavior.
§ 10 · CRO & landing pages
One product. Three
front doors.
- Persona landers — Lonely Evening · Character Builder · Curious Switcher — matched to ad/affiliate creative.
- Plan clarity page — monthly / quarterly / annual + token reality in plain language (what 100 tokens actually buy).
- Create-your-own demo path — reduce time-to-first-character for Persona 02.
- Checkout & trial friction audit — payment methods, discreet billing messaging, trust signals, geo edge cases.
- Experiment backlog — prioritized by expected lift on trial→paid and chargeback risk, not vanity CTR.
§ 11 · Measurement & AI ops
One dashboard.
One weekly decision.
- Source contribution view — affiliate · paid · organic · direct, with trial and paid cohorts separated.
- Creative fatigue signals — concept-level, not just ad-level, so the weekly board is real.
- LTV / refund / chargeback monitors — growth that ignores payment quality is fake growth.
- Policy incident log — what got rejected, why, and the approved rewrite pattern.
- AI-assisted production — faster concept variants and copy banks; human lock on claims and brand-safety.
§ 12 · Engagement shapes
Three shapes.
One recommendation.
Every AYMI engagement is monthly. Media spend, software, affiliate commissions, and creator fees are pass-through and separate from the retainer. The middle tier is starred for a scaled subscription brand already past product-market fit.
Shape 01 · Core
Foundation
Team1 growth strategist
AI dashboardCore operating view
InclusionsBrand-safe creative system · primary lander CRO · trial lifecycle build · monthly review
Best fit if Candy wants a focused infrastructure sprint before expanding paid budget — not the default for a category leader already at scale.
Shape 02 · Scaling
Growth System
Team1 strategist + paid acquisition lead
AI dashboardFull attribution + creative fatigue signals
InclusionsEverything in Foundation + multi-channel paid (policy-aware) + lifecycle expansion + affiliate coexistence model + weekly kill/scale board
Best fit for a revenue-proven subscription brand that needs diversified acquisition and LTV defense without abandoning the affiliate engine that already works.
Shape 03 · Category system
Full Revenue OS
Team2 strategists + executive authority engine
AI dashboardFull stack + custom modeling
InclusionsEverything in Growth System + dual-register brand system + PR/earned + international expansion playbooks + advanced retention modeling
Best fit when Candy is explicitly chasing multi-year category leadership and wants growth + brand authority as one system.
The investment for each shape is held for the scoping call — we'd rather decide together what's actually in scope first, then price it once the answer is real.
§ 13 · Our recommendation
Start with Growth System.
Expand on real data.
Candy.ai is past the “prove anyone will pay” stage. Foundation is the wrong default for a brand already operating as a category commercial leader — it under-scopes paid architecture, multi-channel coexistence, and LTV systems. Growth System is the honest shape for year one: strategist + paid lead, creative cadence, policy playbook, lifecycle spine, CRO on persona landers, and a dashboard that shows whether diversification is actually working.
Full Revenue OS is the graduation path when international expansion, dual-register brand work, and earned-media authority are active priorities — not when the first job is to de-risk affiliate concentration and stabilize paid.
What we need on the call to lock scope: current source mix (% paid / affiliate / organic), trial→paid and month-2 retention, token attach rate, refund/chargeback rates, ad-account history, and which geos are priority. Without those, any budget recommendation is theater.
§ 14 · The 90-day sprint
What the first ninety days
actually looks like.
Days 1–30 · Diagnose & build
Instrument the truth
- Source-mix & funnel audit (trial → paid → M2)
- Policy & ad-account health review
- Persona map + brand-safe creative brief
- First 6 concept frames + 2 landers live
- Trial lifecycle v1 in ESP
Days 31–60 · Launch & learn
Run the board
- Paid tests live under policy playbook
- Weekly kill/scale creative board
- Affiliate coexistence rules (geo / keyword / creative)
- Dunning + win-back live
- Dashboard v1 for weekly decisions
Days 61–90 · Scale what works
Compound
- Scale winning concepts; retire losers
- Second channel or geo expansion if CAC holds
- Token attach & annual upgrade experiments
- Q2 roadmap + budget recommendation
- Graduation criteria to Full Revenue OS
§ 15 · Proof · subscription & recurring consumer
Three brands that share
the mechanical shape.
Candy is not a mattress or a hair vitamin — but the growth physics of subscription consumer brands still apply: creative cadence, LTV defense, and multi-channel acquisition.
Eight Sleep
Why it maps: high-consideration subscription-adjacent consumer brand where creative system + paid discipline moved unit economics — not just volume.
Nutrafol
Why it maps: recurring revenue + retention engine. The Candy job after trial is the same job — keep the months, not just the first charge.
Proven Skincare
+480%
Subscription revenue
Why it maps: personalized digital product + subscription economics. Persona-led creative and conversion infrastructure reduced CAC while growing recurring revenue.
Metrics above are from AYMI’s documented case library. They are adjacent-category proof (subscription / recurring consumer), not AI-companion vertical case studies. We do not invent vertical proof we do not have.
§ 16 · Open items for the scoping call
Six things we want
to hear from you.
01
Source mix today — roughly what % of new paid subscribers come from affiliates vs. paid vs. organic vs. direct?
02
Funnel truth — trial start → paid conversion, month-2 retention, refund/chargeback rates, token attach rate.
03
Ad-account health — which platforms are live, restricted, or dead? Any recent policy incidents?
04
Affiliate terms & concentration — top partners by volume; commission structure; exclusivity or geo rules.
05
Brand boundary — how far public/paid creative can go vs. post-login product reality (the dual-register line).
06
Priority geos & languages — where growth is wanted next vs. where support/compliance is ready.
A compounding
subscription machine.
Candy.ai already proved people will pay for presence. AYMI builds the system that diversifies how they arrive, protects how long they stay, and keeps growth from depending on any single partner or platform.
Next step · 45-min scoping call · studio@aymi.agency