← Back to meeting docs

SFN POC Plan

Formal Proof of Concept Plan · Will evolve into POC PRD
Friday · May 15, 2026

SFN POC Plan

Proof of Concept Plan for the SFN 90-Day Trial Friday May 15, 2026 — v1.0 This document will evolve into the POC PRD (Product Requirements Document) during the trial.

EXECUTIVE SUMMARY

Read this if you read nothing else. Three pages. Everything else is detail.

What we're doing

A 90-day trial that does one of two things: prove SFN can be relaunched profitably, OR prove it can't, decisively, with a documented methodology that ports to Daybreak and beyond. Both outcomes are valuable. The trial is a bet on methodology, not on SFN specifically.

Why this trial exists

SFN is a dormant business with niche-quality content. Peaked at 21,500 monthly sessions in 2022. Generated $5,020 lifetime revenue across ~210 customers (0.11% conversion — 10x below niche benchmark). Founders moved on. 96% organic traffic decline 2023-2025. The content asset is real. The system around it was never built. We're not relaunching a business — we're building the system the business never had.

The DIM Factory POC — what we're proving

One unified factory. Three sub-systems:

  1. Production — AI content factory with multi-model QA pipeline. Ramps from 10-15 quality assets by Day 30 to 60-90 over the full trial. Quality over volume.
  2. Demand Gen — Acquisition + conversion engine. Higgsfield premium creative, lifecycle platform for nurture, automation orchestration.
  3. OS Flywheel — Automation system any tech-savvy operator can run. SOPs, decision logs, agentic workflows. Spins without daily intervention.

DIM, not DARK — humans in the loop, by design. SFN will always need a trained editor for clinical/emotional accuracy validation. Different sub-systems run dimmer than others; production is dimmest, demand gen and OS are half-dim.

Two engines tested in parallel. Four real outcomes.

Demand gen passes Demand gen fails
Content factory passes SCALE — full system works KILL SFN, port factory to Daybreak
Content factory fails RARE KILL CLEAN — no ongoing cost

The content factory is the more portable IP. If demand gen fails but the factory works, the factory transfers. That's a useful outcome, not failure.

Live proof: the agency site case study

24-hour build → optimize loop. Branding AI-generated. Design via Google AI Studio. Deployment via GitHub → Hostinger. Iteration via Claude Code. Optimization via autonomous PageSpeed Insights loop (no human in the inner loop).

PSI scores: 83/84/100/83 → 94/94/96/100. Industry benchmark for top-1000 Google search results: avg mobile 40, avg desktop 60. Only 42% of mobile sites pass Core Web Vitals. The agency site beats the majority of professionally-built websites — built and optimized by AI, supervised at the architectural level only.

Platform recommendations

Course platform: Kajabi + DropInBlog leans recommended ($224/mo combined). The agentic content workflow (via DropInBlog MCP) works equivalently on either Kajabi or Thinkific — that's not the differentiator. Kajabi wins on native funnels, abandoned cart, CRO templates, coaching scheduler. Thinkific wins on webhook granularity, API access, marketplace, 1-click install. Side-by-side test drive confirms (or flips) before Day 1.

Avatar platform: Synthesia is the operator recommendation. Founders decide. HeyGen on paid tiers may produce sharper output (4K), but Synthesia's lip sync and delivery quality is more believable at 1080p. Cost variable: Creator tier ($708-768/yr annual contract) vs Enterprise tier ($12,000+/yr). Sales call after Friday gets firm number.

Landing pages + funnels: Working hypothesis is to build on the same architecture as the agency site (Next.js or plain HTML on Hostinger, agentic workflow optimization). Neither Thinkific nor Kajabi provides good native funnels for our use case.

Budget

Ideal Path: $32,141 Day 1 commitment ($31,433 operating + $708 Synthesia Creator annual) Worst Case Path: $43,433+ Day 1 commitment if Enterprise tier required

Recommendation: Ideal Path. The variable is Synthesia tier.

⚠ Critical: Synthesia annual contracts continue regardless of Day 90 outcome. Even if SFN is killed at Day 90, the remaining 9 months of subscription continue to invoice.

90-day burndown

Foundation (Days 1-14) → Production Proving (Day 30 gate) → Conversion Proving (Day 60 gate + ad spend decision) → Kill/Hold/Scale Decision (Day 90 gate).

Quality over volume — production starts manually, ramps to automation. 10-15 assets by Day 30. 60-90 quality assets cumulative by Day 90.

Three Day 90 outcomes — all planned

Founders' role: advisory, not operational

Founders approve initial budget, strategic direction, wedge product, brand voice, gate decisions. Operator owns system architecture, tool selection, day-to-day execution, new operator handoff documentation, training. Weekly check-ins until ship is sailing in the right direction; then advisory-only at gates. Optional paid editorial role available if founders want deeper involvement.

Scope creep protection

Steve flagged this as the biggest internal risk. Active measures already in place:

  1. Rule of 100 — One primary lane at a time. Order: warm/founder → partner → content → paid ads.
  2. POC vs GTM separated — Content factory and demand gen measured independently. Never blur metrics.
  3. Deferred items have gates — Mobile app: Day 91. Ad spend: Day 60. WordPress migration: Day 90 + 2-year commitment.
  4. Documented in three places — Operator Ownership Agreement, operator manual, this POC PRD. Triple-documented = enforceable.

The ask (Friday meeting)

  1. Course platform direction: Kajabi + DropInBlog leans recommended. Side-by-side test drive confirms (or flips) before Day 1.
  2. Budget approved: Ideal Path $32,141, or honest pushback on what to cut/add.
  3. Ad spend amount: $1K/$2K/$3K at Day 60+. Founders choose.
  4. Gate metrics confirmed for Days 30/60/90.
  5. Operator Ownership Agreement signed.
  6. Synthesia tier: operator calls sales after meeting; firm number for next check-in.

What kills the trial (hard stops)

  1. Founders withdraw commitment (financial or strategic)
  2. Operator cannot continue (health, life, other obligations)
  3. Platform or regulatory issue prevents operation
  4. Clear evidence engines DON'T work (factory produces unusable content, demand gen has zero signal, operator system breaks down)

— END EXECUTIVE SUMMARY —

PART 1 — STRATEGIC FRAME

1.1 Why this trial exists

SFN's history: - Built by Steve and Charlee Roberts. Peaked in 2022 (~21,500 sessions/month). - Generated $5,020 lifetime revenue across ~210 customers. - Founders moved on. Site abandoned by 2023. - 96% Google Organic traffic decline 2023-2025 (Helpful Content Update era). - Despite abandonment, content quality keeps low-volume signups arriving. - One quality post drives 84.5% of historical traffic.

Current state: a dormant business with niche-quality content, real audience traces, and zero active operation. Either it gets relaunched or it gets killed cleanly.

Why now: - Founders have a working hypothesis that AI-driven content + automation can revive it - Operator has the capacity, the agency methodology, and the time - The infrastructure to test it (AI stack, customer lifecycle platform, content factory) is now affordable - 90 days is enough time to know

Why this matters beyond SFN: - If the methodology works, it ports to Daybreak (Steve's other venture) - If it works, it becomes the operator's signature agency offering - If it doesn't work, the playbook documents what doesn't work and why — that's still IP

1.2 Strategic disciplines

Discipline 1: Manual recipe → SOP → assisted handoff → automation Premature automation kills projects. We build manually first, document what works, train an operator, THEN automate.

Discipline 2: Rule of 100 (one primary lane at a time) The order: warm/founder channels → partner channels → content channels → paid ads. We don't run paid traffic at Day 30 just because we can.

Discipline 3: POC and GTM as separate engines The factory (POC) is what we're really proving. Demand gen (GTM) is the test bed for the factory.

Discipline 4: Founders advisory, not operational Founders approve direction, wedge, brand/tone, legal/therapy boundaries, Day 30/60/90 decisions. They don't run the factory.

Discipline 5: Replaceability Every component designed to be swappable. No single point of failure. Phase 2 operator inherits a system, not a personality.

1.3 The three sub-systems of the DIM Factory

1.3.1 Production sub-system

What it is: a multi-model AI content production system that ramps from 10-15 quality assets by Day 30 to 60-90 cumulative quality assets by Day 90.

Success criteria: - Produces 60-90 cumulative quality assets over the trial - Quality passes multi-model QA + human editor review - Operator can run it without the founding operator after handoff - Cost per asset is single-digit dollars

Failure mode: produces volume but quality fails human review, OR produces quality but volume is unsustainable, OR depends on the founding operator's specific skills in ways that can't be handed off.

1.3.2 Demand Gen sub-system

What it is: the applied test of the factory. Takes the content the factory produces and runs it through a funnel: organic discovery → lead magnet → email nurture → wedge product purchase → community/coaching upsell.

Success criteria: - Organic traffic recovery (signs of life, not full restoration) - Lead conversion rate above 1% (10x current 0.11%) - Wedge product validates at $250-500 price point - Email engagement rates match category norms

Failure mode: factory produces good content but no conversion, OR conversion happens but only on paid traffic.

1.3.3 OS Flywheel sub-system

What it is: the documentation, processes, and skills transfer that makes the founding operator replaceable. Manual recipe → SOP → assisted handoff → automation.

Success criteria: - DIM factory SOP documented to the level a trained operator can run it - Therapist HITL editor recruited and trained - Handoff package complete: operator playbook, decision log, voice-of-customer corpus - Founding operator available as paid consultant, not as full-time operator

Failure mode: documentation incomplete, operator candidates can't be found, or processes are too tacit to transfer.

PART 2 — PLATFORM DECISIONS

2.1 Course platform: Kajabi + DropInBlog (leans recommended)

Recommendation flipped twice during research as new information surfaced. Current honest read: agentic capability is equivalent on both platforms (DropInBlog MCP works the same way on either), so the decision comes down to funnel/CRO infrastructure (Kajabi wins) vs operator conveniences (Thinkific wins).

Why Kajabi leans: - Native landing pages + funnels (Thinkific has none worth using) - Native abandoned cart recovery (Thinkific has none) - CRO-optimized templates, coaching scheduler, premium feel - These are expensive to build ourselves and compete with content production for operator time

Why Thinkific is the legitimate alternative: - Lesson-level webhooks (Kajabi fires on purchase events only) - API included at Grow tier (Kajabi gates this to add-on) - Open public app marketplace - DropInBlog 1-click install (Kajabi requires custom code paste)

Test drive required before Day 1 lock: UX feel, theme customization friction, content publishing workflow rhythm, abandoned cart automation quality. Research can't capture these. Both platforms have free trials. Side-by-side test drive resolves what research can't.

Tradeoffs we own honestly: - Both Thinkific and Kajabi are walled gardens — automation logic doesn't export from either - DropInBlog renders posts in JavaScript (some social-share quirks; documented workarounds exist) - Kajabi Sept 2025 price hike was first major increase in 15 years

2.2 Avatar platform: Synthesia (locked)

Three platforms tested on free tier. Same script. - HeyGen: generic output, poor lip sync - Colossyan: worse than HeyGen - Synthesia: decisive winner, production-quality

Tier uncertainty: Creator ($59-64/mo annual = $708-768/yr) vs Enterprise ($1,000+/mo annual = $12,000+/yr). Online info on Enterprise is shabby. Operator calls Synthesia sales after Friday meeting to get firm number.

Creator plan capacity (verified): 360 minutes per year, available all at once (44,000 credits). Fits wedge production at ~5 hours with pre-verified audio workflow. Unused minutes do NOT roll over at renewal.

Studio Avatar add-on: $1,000/year on annual plans. May or may not be required depending on Personal Avatar quality outcome.

Content moderation flag: Multiple user reports of accounts banned for legitimate business content. Pre-test moderation tolerance during onboarding.

⚠ Annual commitment reality: Even if SFN is killed at Day 90, Synthesia annual contract continues for remaining 9 months. This is not 90-day cost; it's annual commitment Day 1.

2.3 Customer lifecycle platform (locked)

Operator's custom CRM. $150/mo flat to SFN. Includes email, SMS, automation, funnels, source attribution, social posting + management, abandoned cart, course abandonment, database reactivation, automation orchestration.

2.4 Premium content suite: Higgsfield (locked)

Already MCP-wired into agentic workflow (Claude Code, Cowork). Annual prepay recommended: $1,200 upfront for full year (50% off monthly rate). Unused credits roll forward to Daybreak. SFN allocation: $300 of $1,200 annual = $100/mo effective for SFN.

2.5 AI stack

2.6 Landing pages and funnels — working hypothesis

Neither Thinkific nor Kajabi gives us great native funnels for our use case. Kajabi is better than Thinkific here but still walled. The working hypothesis is to build marketing landing pages on the same architecture as the Bearded Brothers agency site: Next.js or plain HTML on Hostinger, deployment via GitHub, optimization via agentic workflow.

Why this works: - Same stack we already proved overnight (94/94/96/100 PSI scores beats SFN's current Wix site by 41 points on performance) - Full ownership of funnel logic — no walled-garden lock-in - DropInBlog MCP integrates regardless of host - Agentic workflow can iterate landing pages the same way it iterated the agency site - Cost-effective vs paying Kajabi premium for landing page features

Final architecture (Next.js vs plain HTML vs Astro) decided after deeper research. This is the lean-toward, not yet locked.

Replaceability: same as everything else in the stack — replaceable by another framework/host combination if needed.

2.7 Multi-model QA — terminology and pattern

What it does: When the AI content factory produces an asset, that asset is checked by QA agents running on different models before human editor review. Each model has different strengths and blind spots; cross-model validation catches errors any single model would miss.

Models in the stack: - Claude (Anthropic) — primary content reasoning, strategic logic checks - GPT (OpenAI) — alternate perspective, broad-knowledge fact-checking - DeepSeek — code/structural validation, cost-efficient bulk QA - Grok — current-events fact-checking, real-time data validation - Gemini (Google) — search-grounded validation, source verification

Why this is the DIM Factory pattern: humans don't read every word of every asset before publication. The system catches issues. Editor reviews flagged items only. This is what "dim" looks like in content production — most of the QA work happens autonomously; the editor is the human-in-the-loop intervention point for edge cases.

PART 3 — BURNDOWN PLAN

3.1 Phase track

Phase Days Focus
Phase 01 1–14 Foundation: platform lock, avatar pipeline, voice clone, wedge candidate, DIM Factory v0 manual recipe
Phase 02 15–30 Production Ramp: first 10-15 quality assets, multi-model QA pipeline running, funnel live, wedge product build initiated
Phase 03 31–60 Conversion Proof: wedge product launched, first customers, ad spend decision at Day 60 gate
Phase 04 61–90 Scale or Decide: conversion metrics matured, operator handoff package complete, Day 90 decision

3.2 Gate metrics

Day 30 — Production proving

Day 60 — Conversion proving + Ad spend decision

Day 90 — Kill / Hold / Scale

3.3 Three Day 90 outcomes

Outcome Trigger Cost
Kill Revenue floor missed by 50%+, OR conversion <0.5%, OR factory unable to produce sustainably $0 wind-down + Synthesia annual continues
Hold Revenue floor met but trajectory unclear, OR new wedge candidate worth testing ~$8,500/mo
Scale Revenue floor met + conversion ≥1.5% + factory producing sustainably + operator system handoff-ready ~$90K Year 1

PART 4 — OPERATOR & FOUNDER GOVERNANCE

4.1 Founders' role

Founders are advisory, not operational. Weekly check-ins during early weeks until the ship is sailing in the right direction; then advisory-only at gates.

Founders approve: - Initial budget and any budget changes - Strategic direction at kickoff and gate reviews - Wedge product selection - Brand voice, tone, visual identity - Legal and therapeutic content boundaries - Day 30/60/90 gate decisions - Ad spend amount at Day 60

Founders do NOT: - Produce content - Operate the customer lifecycle platform or course platform - Approve individual content pieces - Attend weekly check-ins (gate reviews only after ship is sailing)

Optional paid role: Founders may take on Editorial Director or similar paid role if they want deeper involvement.

4.2 Operator ownership

Operator owns, without seeking approval: - System architecture - Tool selection within approved stack - Content production volume, cadence, topic prioritization - DIM Factory architecture and methodology choices - Editor recruitment, training, management - Day-to-day execution and operations - New operator handoff documentation and training - Multi-model QA pipeline composition

Operator surfaces to founders with proposed direction: - Wedge candidates with recommendation - Course platform finalist - Synthesia tier decision after sales call - Budget overruns approaching the 10% buffer - Any blocker that threatens gate metrics

PART 5 — SCOPE CREEP PROTECTION

5.1 Why this matters

A 90-day trial dies when teams optimize the wrong thing, add the wrong tool, or build the wrong feature on Day 45. Scope creep is the biggest internal risk to the trial. Active measures are in place.

5.2 The four protections

Rule of 100 — One primary lane at a time

Order: warm/founder channels → partner channels → content channels → paid ads. Each lane proves before the next activates.

Example: We don't run paid traffic at Day 30 just because we can. Organic conversion must prove first. Day 60 gate is where paid traffic gets approved.

POC vs GTM separated — Never blur metrics

Content factory (POC) is the IP we're proving. Demand gen (GTM) is the test bed.

Example: If content factory ships 60-90 quality assets but GTM fails to convert, that's NOT failure — it's a useful outcome. Factory ports to Daybreak. Don't blur "content quality" with "conversion rate" in reporting.

Deferred items have gates — Nothing added mid-trial

Mobile app: Day 91 pre-commit. Ad spend: Day 60. WordPress migration: Day 90 + 2-year commitment.

Example: If a founder says "let's add an Instagram strategy on Day 45" — the answer is "great, that goes in the Phase 2 parking lot, not the current trial." No exceptions, even for good ideas.

Documented in three places — Not slideware

Scope creep protection lives in the Operator Ownership Agreement, the operator manual, and this POC PRD. Triple-documented = enforceable, not aspirational.

Example: When founder or operator wants to add scope, the answer references the doc, not the slide. "Per Section 5 of the POC PRD, this goes to the parking lot for Phase 2 evaluation."

5.3 Parking lot — Phase 2 ideas

Items that come up during the trial that don't kill the trial but warrant Phase 2 consideration go here. Examples (placeholder for now): - Mobile app activation (Day 91+ if validated) - Open-source platform migration (BuddyBoss + LearnDash) - Additional content verticals - Community/cohort offering - B2B partnerships - Affiliate program - Podcast launch

PART 6 — RISKS

Risk Mitigation
AI content flagged as low-quality by Google Multi-model QA pipeline + HITL editor for clinical accuracy
Wedge product selection wrong Voice-of-customer-driven research, validated before build, single primary lane
Synthesia Enterprise expense unknown No real mitigation. Sales call after Friday gets firm number. Founders approve actual cost.
Synthesia content moderation strict Pre-test moderation tolerance during onboarding
Mobile app deferred — perceived as cut scope Calendar reality: 6-8 weeks review + 8-12 weeks build = couldn't ship before Day 90. Pre-commit to Day 91 activation.
Platform choice locked in error Parallel side-by-side trial before next meeting. Lock decision before Day 1.
Paid traffic doesn't scale Deferred to Day 60 organic conversion proof
Speaker rights on founder archive Mitigated by citing sources in all derivative content

PART 7 — OPEN QUESTIONS

7.1 Resolved Friday

  1. Course platform direction (Kajabi + DropInBlog leans recommended; test drive confirms before Day 1)
  2. Budget approval level (Ideal Path $32,141 or Worst Case $43,433+)
  3. Ad spend amount selection
  4. Day 30/60/90 gate metric confirmation
  5. Operator Ownership Agreement signed

7.2 Resolved post-Friday

  1. Synthesia tier (Creator vs Enterprise — sales call decides)
  2. Wedge product final selection (Tuesday Semrush data → decision)
  3. Course platform final lock (after side-by-side test drive)
  4. Editor candidate identified

7.3 Resolved during trial

  1. Day 30 customer interview activation (or not)
  2. Day 60 ad spend activation amount
  3. Community vs no-community direction (pending EB research)
  4. Day 90+ migration to open-source decision

PART 8 — DOCUMENT EVOLUTION

This POC Plan v1 is the formal scope document for the 90-day trial. It will evolve:

The Operator Ownership Agreement governs operator/founder interaction. The Operator Manual contains technical SOPs. This POC PRD contains strategic scope.

POC Plan v1 · Friday May 15, 2026 · Owner: Operator