User Journey¶

Phase: 6 — Product Project: likeness Date: 2026-05-09 Confidence: Medium — journey logic is grounded in research and existing mockup; specific timing and emotional arcs need validation against real creator and fan behavior in concierge.

Two journeys, intersecting¶

Likeness has two primary user types. Their journeys are independent — a creator can complete the full creator journey before any fan sees them — but they intersect at the moment a fan generates content and the creator earns from it. That moment is the heart of the product.

CREATOR JOURNEY FAN JOURNEY ─────────────── ────────── 1. Discovery 1. Discovery (via creator) 2. Application 2. Subscription 3. Verification 3. AI tier exploration 4. Onboarding (the disclosure) 4. First generation 5. Source upload + curation ←─ INTERSECTION ─→ 6. License configuration 5. Submission to creator 7. Model training 6. PPV unlock 8. First test generations 7. Ongoing 9. Soft launch to existing fans 10. First fan-driven generation 11. First payout 12. Ongoing operations

Creator Journey¶

Stage 1: Discovery — "I heard about Likeness"¶

How they arrive: - Founder or Creator Ops cofounder reaches out 1:1 (the primary path at MVP) - Industry-event introduction - Peer creator referral (post-concierge)

What they're feeling: - Default skepticism. "Another platform pitching me." - Curiosity if the consent angle is named immediately.

What the platform should do: - The first contact should not feel like a sales pitch. It should feel like a colleague calling about something specific. (See the cold-outreach voice register in tone-of-voice.md.) - Lead with the problem ("you've probably had AI of you made without consent") and the architectural commitment ("the model never leaves the platform"). Not the features, not the pricing.

Drop-off risk: High. Most creators will say no on the first call simply because new platforms get a no by default.

Mitigation: Don't push closure on call one. The follow-up is a 30-minute conversation walking through the mockup and the license configuration UX. That's where curiosity becomes interest.

Stage 2: Application — "I'd like to participate in concierge"¶

Trigger: Creator agrees to the second conversation. The application is light — name, business info, existing platform stack, audience size, basic bio.

What the platform should do: - Do not gatekeep applications heavily. The concierge cohort is invite-only by design — Creator Ops decides who advances. - Make the application form lightweight (5 minutes). Heavy gates here lose people who would have been good fits.

Drop-off risk: Low at this stage if they've made it past stage 1.

Stage 3: Verification — "Prove you are who you say you are"¶

What happens: - Government ID upload + liveness check via third-party provider (Persona / Yoti / Veriff) - 2257 record creation begins - Background check on past content history if relevant (concierge case-by-case)

What the creator is feeling: - Mild friction. They've done identity verification on other platforms; this isn't surprising. - Watching for signals of how the platform handles their data.

What the platform should do: - Be specific about why each piece of information is required (regulatory, processor, 2257). The voice from tone-of-voice.md applies directly. - Hold ID uploads in encrypted storage with strict access controls; surface that posture in the UX.

Drop-off risk: Low-medium. Some creators may decline if the diligence feels excessive. Calibrate to the 2257 / Mastercard floor — no more, no less.

Stage 4: Onboarding — "Three things to know before you train your model"¶

This is a load-bearing brand moment. The creator has been verified and is about to commit time to training. Before anything else, they see the disclosure (sample copy in tone-of-voice.md, "Onboarding screen" sample):

We will never export your model.

Every output is watermarked, signed with your license ID, traceable to the buyer.

You can revoke at any time. Files already downloaded are still on people's devices — we can't change that, but we can help you take them down.

Why this matters: - It's the platform's last clear opportunity to surface dealbreakers BEFORE the creator invests time in source material curation and model training. - It demonstrates the brand's "Plain Talk" value (see mission-vision-values.md). - A creator who continues past this screen is operating on full information and can't reasonably claim later that they were misled. That's both ethical and operationally protective.

Drop-off risk: Some creators will pause here. A few will leave. Both outcomes are correct. The ones who continue are the ones who actually want this product.

Stage 5: Source upload + curation¶

What happens: - Creator uploads 30-100 curated source images (per ML brief) - Platform pre-publication-reviews each image (T&S, automated + human) - Creator selects which images to include in training data

What the creator is feeling: - This is real work. Source curation is non-trivial — they're choosing which version of themselves the AI will reflect. - Concerned about which photos to include. Wants to get this right.

What the platform should do: - Provide guidance: variety of angles, expressions, settings; consistent quality. (Creator Ops provides 1:1 support during concierge.) - Be explicit about what the model will and won't be able to generate based on input variation. Set realistic expectations.

Drop-off risk: Medium. Some creators discover they don't have 30+ usable source images, or aren't comfortable curating. Concierge support compensates.

Stage 6: License configuration — "Set the rules"¶

This is the most differentiated UX in the product. The creator works through the license object: allowed categories, blocked categories, distribution rules, per-fan permissions.

What the creator is feeling: - Initially overwhelmed. Some of these options are things they've never had to articulate before. - After 10-15 minutes, increasingly empowered. They are deciding how their AI gets used. Most platforms don't give them this.

What the platform should do: - Provide sensible defaults that approximate "what most creators choose." Defaults reduce paralysis. - Surface how other (anonymized) concierge creators have configured similar fields. Show the range, not just a static default. - Explain each option in plain language, with concrete examples. - Make changing the license easy and forward-effective. The license object is not a contract — it is a set of rules that updates whenever the creator wants. Conveying this is essential.

Drop-off risk: Low if defaults are good. Medium if the UX feels like a legal form.

Stage 7: Model training¶

What happens: - ML Lead and team train the per-creator LoRA + face adapter (concierge phase: this is high-touch). - Creator does not directly interact during training; they wait. - Estimated training time: 24-72 hours per creator.

What the creator is feeling: - Anticipation. This is the moment where the AI version of them comes into existence.

What the platform should do: - Communicate progress. "Training in progress, expected completion Friday." Keep them updated, set expectations. - During concierge, Creator Ops checks in personally.

Drop-off risk: Low at this stage. The creator has invested too much to leave without seeing the output.

Stage 8: First test generations¶

What happens: - Creator gets first access to test generations. They prompt the model themselves (within their own license rules) and see results. - Creator and ML team iterate if quality is below bar.

What the creator is feeling: - This is the first major emotional moment of the journey. If the AI looks like them, recognizably them, the product becomes real for them. If it doesn't, the project pauses for re-training.

What the platform should do: - Set quality expectations honestly. Some creators with limited source material will get less faithful models. Don't oversell. - Provide tooling for creators to easily flag quality issues so ML Lead can iterate.

Drop-off risk: Medium. If the model looks bad, this is where creators leave or ask for a refund of their time.

Stage 9: Soft launch to existing fans¶

What happens: - Creator announces AI tier to their existing fan audience via their normal channels (their existing platform, email list, social). - Likeness provides templated launch copy that the creator can customize. - Some fans subscribe immediately; most wait and watch.

What the creator is feeling: - A version of stage fright. They are publicly committing to this experiment. - Watching the first signups.

What the platform should do: - Provide visible analytics: how many fans subscribed, on what tier, in what time frame. - Surface signals about which fans are existing high-value subscribers and which are new — this informs the creator's outreach.

Drop-off risk: Low for the creator (they've gone too far to leave); high for fans (most won't convert immediately).

Stage 10: First fan-driven generation — THE AHA MOMENT¶

What happens: - A fan, paying real money, generates their first piece of content using the creator's AI model. - The creator sees it in their dashboard. License rules were enforced. The output is watermarked. The fan paid.

What the creator is feeling: - The aha moment. This is the first time the system delivers on the value proposition end-to-end. The creator earned money on AI of themselves, on their rules, with provenance. - Often: surprise that it actually worked.

What the platform should do: - Make this moment visible. The creator dashboard should highlight the first generation by a fan, the first revenue, the first audit log entry where the license was enforced. - This is where Creator Ops checks in personally during concierge — to share the moment, hear the creator's reaction, and capture testimony.

Drop-off risk: None. This is when retention becomes real.

Stage 11: First payout¶

What happens: - First creator payout flows through the multi-processor stack to the creator's bank account. - Reconciliation report shows gross, platform fees, processor fees, net.

What the creator is feeling: - Trust crystallizing. The money is in the bank. The platform did what it said.

What the platform should do: - Make the reconciliation transparent. Show every fee. Don't hide the processor cut. - Use this moment to begin asking for peer referrals. Creators trust other creators after they see the money work.

Drop-off risk: Low.

Stage 12: Ongoing operations¶

What this looks like: - Creator regularly reviews the approval queue (fan submissions for review) - Creator updates their license rules based on patterns they see - Creator earns revenue across the mix (subscription floor + AI generation credits + submission fees + PPV) - Creator interacts with Creator Ops for issues, optimizations, peer connections

What the platform should do: - Surface optimization opportunities (e.g., "your $50 tier has 80% conversion from the $25 tier — consider promoting it more") - Maintain a regular cadence of Creator Ops touchpoints during concierge phase - Begin exposing the creator to other concierge creators for peer learning, with explicit consent

Drop-off risk: Real. Concierge creators may churn at month 3-6 if the AI revenue line doesn't grow or if their fans don't sustain engagement.

Fan Journey¶

Stage 1: Discovery (via creator)¶

How they arrive: - The creator they already follow announces the AI tier on the creator's existing channels (OnlyFans / Fansly / direct). - The creator is the discovery mechanism. Likeness is not a search platform.

What they're feeling: - Curiosity, sometimes confusion ("AI of [creator]? Like a deepfake?"). The creator's framing matters. - Some fans are immediately interested; most are skeptical.

What the platform should do: - Provide creators with high-quality launch templates that explain the consent and licensing posture. Fans need to understand "this is the creator's licensed AI, not a deepfake" within the first 60 seconds of arriving. - A landing page anchored to the creator (not the platform) — the creator's visual brand, with Likeness as the engine.

Drop-off risk: Very high at this stage. Most fans will visit and leave.

Stage 2: Subscription¶

What happens: - Fan picks a tier ($15-$200 illustrative), enters payment info via processor stack, completes age verification, gets account access.

What the fan is feeling: - Standard creator-platform subscription experience. They've done this before. - Mild friction at age verification (US baseline) — not heavy.

What the platform should do: - Match the speed and feel of OnlyFans / Fansly subscription flows. Don't add friction that wouldn't be there on the substitute platform. - Surface what each tier includes specifically, including AI generation credit allocation if applicable.

Drop-off risk: Medium. Standard funnel drop-off, mostly comparable to existing creator platforms.

Stage 3: AI tier exploration¶

What happens: - Fan sees the AI gallery (creator-approved generations) on their tier - Fan reads about generation credits and per-generation costs - Fan considers whether to upgrade tier or buy additional credits

What the fan is feeling: - Comparing value vs. existing AI girlfriend platforms (Candy.ai, etc.) and vs. real content from this creator. - Starting to imagine what they'd generate.

What the platform should do: - Make the gallery a strong showcase. Quality generations approved by the creator. - Show the license rules in plain language so fans know what they can and can't ask for. Predictable, not punishing. - Make pricing transparent: this many credits = this many images at this rate.

Drop-off risk: Medium-high if the gallery is weak or the pricing feels confusing.

Stage 4: First generation — THE FAN AHA MOMENT¶

What happens: - Fan writes their first prompt - Prompt parses against the license — either approved (continues to inference) or rejected (returns explanation) - If approved, generation runs and an output appears - Output is watermarked and added to the fan's private collection

What the fan is feeling: - The aha moment. This is when "AI of [creator]" stops being abstract and becomes concrete. If it looks like the creator, the fan has experienced what they paid for.

What the platform should do: - Make the first generation feel like a small event. The creator's brand on the page. The output presented well, not just dumped. - Make it easy for the fan to take the next step (generate more, submit for creator approval, save to their collection, share with the creator privately).

Drop-off risk: None at this stage if the output is good. Output quality is everything.

Stage 5: Submission to creator¶

What happens: - Fan likes a generation, decides to submit it for creator review - Pays a submission fee - Goes into the creator's approval queue

What the fan is feeling: - Engagement deepens. The fan is now interacting with the creator (asynchronously) about a piece of content they made together.

What the platform should do: - Make submission feel like a conversation, not a transaction. The fan can include a note. The creator can respond. - Set realistic expectations on response time (concierge: 1-3 days typical).

Drop-off risk: Low for engaged fans; high as a percentage of all fans (most won't submit).

Stage 6: PPV unlock — sometimes¶

What happens: - Creator approves a generation and adds it to a public gallery as PPV - Other fans (or the original submitter, if they want a higher-resolution / unwatermarked-on-fan-side version) can unlock it - Submission-fee + PPV-unlock split goes to creator + platform

What's important: - This is where the OnlyFans-style 59%-one-off revenue dynamic emerges. PPV unlocks compound on the AI generation revenue line.

Drop-off risk: Low — this is a pattern fans already understand from existing creator platforms.

Stage 7: Ongoing¶

What this looks like: - Recurring subscription with monthly credit allocation - Occasional credit top-ups - Regular generation and occasional submission - Engagement with the creator's gallery and approved-content feed

What the platform should do: - Track usage patterns; surface to the creator (with fan consent) so they can engage their best fans more. - Watch for retention drop-off at month 1, month 3, month 6 — these are standard creator-platform churn points.

Drop-off risk: Standard creator-platform churn. Adult subscription churn is high (often 40-60% monthly). The mitigation is high lifetime value at the active fans, not retention of all signups.

The Aha Moments — summary¶

Persona	Aha moment	Why it matters
Creator	First fan-driven generation: a fan paid real money, ran a prompt, the license was enforced, the output was watermarked, and the creator earned.	This is when the entire value proposition — consent + revenue + control — collapses into a single observable event.
Fan	First generation that looks recognizably like the creator and was generated within rules the creator set.	This is when "AI of [creator]" stops being a concept and becomes content. Output quality is the deciding factor.

The product should be designed to make these moments arrive as quickly as possible without skipping the trust-building work that precedes them. Specifically:

For creators, the gap between subscribing and the first fan generation is the period of highest churn risk. Concierge ops should drive a creator to first-fan-generation within 30 days of training completion.
For fans, the gap between first subscription and first generation is the period of highest churn risk. The UX should encourage a first generation within minutes of subscribing.

Drop-off risk summary across both journeys¶

Stage	Risk level	Mitigation
Creator first-call	High	Lead with the architecture commitment, not features
Creator onboarding disclosure	Medium	This is by-design filtering; don't try to suppress drop-off
Creator source curation	Medium	Concierge support is the answer
Creator first test generation (quality bad)	Medium	ML iteration; honest expectations upfront
Creator soft launch (low fan conversion)	High	Set expectations: most fans don't convert immediately
Fan first visit (no context)	Very High	Creator's launch copy is what manages this
Fan subscription funnel	Medium	Match competitor frictionlessness
Fan first generation (quality bad)	High	ML quality is the platform's responsibility

Strategic Connections¶

The aha moments map directly to the success criteria in mvp-definition.md — first fan generation is the activation metric; per-fan AI spend is the revenue metric.
The trust-building stages (3, 4, 6) reflect the brand personality work in brand-personality.md ("trust under skepticism" as the primary emotion).
The drop-off risk analysis informs which features in feature-prioritization.md need extra UX investment — onboarding disclosure (#11), license configuration (#1), and fan generation interface (#14).
The two-journey intersection at first fan generation aligns with the financial sensitivity analysis in 05-financial/revenue-model.md — this single event is what the model is testing.

Flags¶

Red Flags: - The creator journey is long. From first contact to first revenue is realistically 6-10 weeks during concierge. This must be communicated upfront. Creators who expect "instant onboarding" will churn.

Yellow Flags: - The fan-side aha moment depends entirely on AI output quality vs. the recognizable creator. If ML quality is below bar at concierge launch, fans churn at the moment the platform was supposed to convince them. ML Lead's quality-evaluation methodology (per ML brief) needs to set a floor before fans see anything. - The "Stage 9: Soft launch to fans" stage is where concierge phase risks visible failure. If a creator launches and few fans convert, that's public-to-the-creator's-audience. Concierge support during this phase is critical.

Sources¶

mvp-definition.md — feature scope ground truth
feature-prioritization.md — UX investment grounding
brand-personality.md — emotional target
tone-of-voice.md — voice for each journey moment
01-discovery/target-audience.md — persona basis
docs/founder-brief.md and existing mockup/ — original journey logic this document formalizes