Revenue Model¶
Phase: 7 — Financial (Fast Track: revenue model only; full projections deferred until concierge data exists) Project: likeness Date: 2026-05-09 Confidence: Medium on benchmarks; Low on Likeness-specific projections (no concierge data yet)
Honest framing for this phase¶
Under Fast Track, full 12-month / 3-year projections are NOT produced because they would be fabricated. Likeness has no creator participation data, no actual processor terms, and no real-world inference cost data. This document focuses on the revenue model shape, the unit economics floor, and the assumptions to test in concierge — those are the inputs to a real financial model later.
If the concierge phase produces 6 months of real per-creator data, full projections become writable. Until then, every line below is either an industry-benchmark anchor or a clearly-labeled estimate.
Pricing strategy¶
Subscriber tiers (creator-set, illustrative)¶
Per the founder brief — confirmed by research as benchmark-reasonable:
| Tier | Monthly | What it includes |
|---|---|---|
| Real content only | $15 | Real photos / videos / messaging |
| Real + AI gallery viewing | $25 | Approved AI gallery access |
| Real + AI generation credits | $50 | Monthly AI generation credit allocation |
| Premium with submissions | $100 | Higher credit allocation + submission privileges |
| Self-insert (post-MVP) | $200 | Fan trains a likeness model of self; collab generations |
These are anchored to existing OnlyFans tier ranges ($4.99-$49.99 typical, $50-$100 for top creators). Each creator sets their own tier prices.
Per-generation pricing (compute credits)¶
Adult AI generators charge $0.20-$0.50 per image (e.g., Candy.ai's $0.40/image at the $9.99/100 tokens rate). Likeness's pass-through cost + creator markup model should land in the $0.30-$1.00/image range depending on creator-set markup.
Operating principle: the platform passes through real inference cost + a small platform margin (5-10%); the creator sets the markup on top. Creator typically captures 60-80% of the per-generation revenue.
Submission & approval queue fees¶
Creator-set, typically $1-$10 per submission. Creator captures 70-80%; platform takes the standard ~20% rate.
PPV unlocks of approved generations¶
Standard creator economy mechanic. Creator sets price per unlock. ~20% platform take.
Revenue mix model¶
Anchored to the OnlyFans benchmark of 59% one-off vs. 41% subscription:
| Revenue type | Likeness target % at maturity | Notes |
|---|---|---|
| Subscription (across all tiers) | 25-35% | Floor revenue |
| AI generation credits | 30-40% | The new primary revenue line |
| Submission fees | 10-15% | Creator-controlled scarcity |
| PPV unlocks | 15-20% | Standard creator economy |
| Custom requests / messaging | 5-10% | Carry-over |
This is the target mix at maturity. At concierge launch, it'll be subscription-heavy (60-70% subscription) until fans adopt AI generation. The trajectory toward this mix over Year 1 is itself a meaningful indicator.
Platform take rate analysis¶
Industry standard: OnlyFans takes 20%. Likeness should match this — going higher creates creator-trust friction; going lower compresses platform margin in a category that already has higher cost structure (adult-friendly processor fees, compliance overhead).
Net to creator after Likeness's 20% + processor fees (5-15%): Creator nets ~65-75% of gross revenue, vs. ~78% on OnlyFans. This is the gap to address in creator outreach. Creator outreach should not dodge the question; it should explain that the difference funds the consent infrastructure (no model export, license-gated inference, watermarking, takedown ops, Compliance Lead at $115K cash) that creators don't currently get from OnlyFans.
Unit economics anchors¶
Per active fan (subscriber)¶
Anchored to OnlyFans benchmarks:
- Avg fan ARPU on OnlyFans 2024: ~$19/month (derived from $7.22B / 377M fans / 12 months — though many fan accounts are inactive)
- Active fan ARPU on top creators: $50-$200/month typical
- Likeness target active-fan ARPU at maturity: $40-$120/month, with the AI-generation tier driving the upside
Per active creator¶
- Likeness platform revenue per active creator at maturity: [Estimate] $400-$2,000/month
- Wide variance by creator size and fan AI-adoption rate
Margin profile¶
- Adult-friendly processor fees: 5-15% of GMV (vs. 2-3% mainstream)
- Identity verification per onboarding fee: [Estimate] $1-$5
- Inference compute pass-through: zero net margin (passed through with small markup)
- Platform-side inference (training, internal R&D, free-tier overhead): real cost, scales with creator count
- Personnel + compliance overhead: $940K + $165K = $1.105M annual at MVP; constant in early phase
Realistic platform-level gross margin at maturity: 45-65% depending on processor mix, inference volume, creator mix.
Sensitivity analysis (what changes the picture)¶
Variables that move the model materially:
| Variable | Pessimistic | Base | Optimistic |
|---|---|---|---|
| Concierge creators retained at month 6 | 30% | 60% | 80% |
| % of fans on AI-generation tier within 6 months | 5% | 15% | 30% |
| Per-active-fan AI-credit spend / month | $20 | $50 | $150 |
| Processor fees | 15% | 9% | 5% |
| Year 1 platform revenue | $40K | $200K | $1M+ |
The most leveraged variable: % of fans on AI-generation tier. If Likeness only gets 5% of fans onto the $50+ tier, the AI revenue line is small enough that Likeness is essentially an OnlyFans-like platform with a takedown component — workable but not differentiated. If 30% of fans adopt, the AI revenue line dominates and the unit economics work.
This is the testable question for concierge.
Path to Series A unit economics¶
Based on Loti / Vermillio / Fanvue comparables, a credible Series A in this space is $10-20M at $3-8M ARR with growth + clear expansion narrative.
Likeness's path: - Year 1 (concierge): $40-$200K platform revenue from 5-10 creators. Goal: prove unit economics shape (revenue mix, margin, retention). - Year 2 (closed beta): $2-5M ARR from 500-3,000 creators. Goal: prove scaling motion. - Year 3 (early scale): $10-20M ARR + voice/video expansion narrative. Series A target.
Each transition depends on the previous validation succeeding. Pre-seed → seed bridge typically requires either (a) clear concierge unit economics or (b) bridge round to extend runway through closed beta.
Funding context¶
- Pre-seed target: $1.5M for 18 months (per
docs/budget.md). - Personnel + legal + infra + compliance + creator acquisition + office + reserve allocation per founder budget.
- Seed-ready milestones: structured concierge unit economics + processor relationships + initial creator-side referral motion.
- Bridge risk: if processor onboarding delays by 3+ months, runway runs without revenue. The reserve absorbs one bad month; multi-month delay requires a bridge round.
Key Assumptions to Test¶
These are the inferred-not-validated assumptions whose truth or falsity drives the model:
| # | Assumption | Validation method |
|---|---|---|
| 1 | Adult creators with existing audiences will participate at the proposed economics (20% take + processor fees) | 10-15 structured creator interviews; concierge cohort signing |
| 2 | Fans of those creators will adopt AI-generation tiers at >15% of subscribers | Concierge cohort A/B test |
| 3 | Per-fan AI-credit spend lands at $50+/month | Concierge data |
| 4 | Adult-friendly processors approve Likeness on consent-posture diligence | Pre-launch processor BD |
| 5 | Per-creator LoRA training and inference cost stays within passed-through markup | ML Lead + concierge data |
| 6 | Creator concerns about fan-trust erosion don't suppress fan AI adoption | Creator interviews + concierge fan retention metrics |
Data Gaps¶
- No actual processor terms. All processor-fee assumptions are anchored to industry ranges, not signed terms.
- No creator-specific WTP data. All creator-economics math rests on inferred preferences.
- No actual inference cost data per LoRA + face-adapter + ControlNet stack. ML brief flags this; ML Lead must own the curve.
- No fan AI-tier adoption data. Fanvue's 15% AI revenue share is the closest benchmark; Likeness's specific adoption rate is unvalidated.
Strategic Connections¶
- The 59%-one-off-vs-41%-subscription mix from
01-discovery/raw/customers-demand.mddirectly drives the revenue mix targets above. - Processor fee assumptions reflect the regulatory analysis in
01-discovery/raw/regulatory.md(Mastercard / Visa rules, adult-friendly processor landscape). - Sensitivity analysis variables align with the validation experiments in
06-validation/.
Flags¶
Red Flags: - The model is honest about being a directional shape, not a financial forecast. No specific Year 1 revenue number should be quoted to investors as a forecast — only as a range with explicit assumption labels.
Yellow Flags: - The 65-75% net-to-creator after platform take + processor fees creates a real "vs. OnlyFans" objection. Address head-on with creator outreach. - The fan AI-tier adoption rate is the most leveraged variable in the model. It's also the hardest to predict pre-launch. Concierge cohort design should isolate this variable cleanly.
Sources¶
docs/budget.md— capital plan01-discovery/market-analysis.md— unit economics benchmarks01-discovery/raw/customers-demand.md— pricing intelligence- Industry benchmarks (Tier 2/3): OnlyFans 80% creator share, Candy.ai per-image pricing, adult-friendly processor fee ranges