Skip to main content
Synced from the repo — do not edit here

Canonical source: docs/claude/business-strategy.md. This page is generated by docs/scripts/sync-handbook.mjs. Edit the source file in the repo; changes appear here on the next build.

Business Strategy — Pricing Rationale, Margins & Scale

Captures the strategic reasoning behind the pricing model. For the live pricing numbers see docs/claude/billing-pricing.md and root CLAUDE.md. Interactive artifacts on Desktop: ready-practice-strategy-memo.html, ready-practice-arr-model.html ($10M), ready-practice-100m-model.html (platform), ready-practice-pricing-sheet.html, ready-practice-quote-builder.html.

1. The core decision: cash-cow, not venture land-grab​

Chosen path: premium, profitable, owner-controlled — charge up front, no free tier, ~80% margin, ~75% ownership retained. NOT a VC-funded GMV platform.

Reasoning:

  • No conviction in a $1B ceiling. Without belief in the generational outcome, the venture treadmill is a bad trade (dilute to ~12–18% chasing an outcome you don't believe in).
  • High-probability $75M beats low-probability $200M. Risk-adjusted, keeping control of a likely outcome wins. 18% of a maybe-$700M vs 75% of a likely-$100M.
  • Margin is the owner's moat. 80% gross-margin software throws off cash; GMV/payments revenue is high-headline, low-margin — great for fundraising decks, worse for the wallet.
  • Build-profitably-first forecloses nothing. Reaching ~$10M with little dilution earns the right to either keep milking it or raise later on the platform thesis from strength (with real GMV data). Raising big now is the only irreversible move.

Guardrails (how the cash cow dies): coasting (the exit multiple is bought with growth, not margin — a flat cash cow sells 4–6× revenue, a growing one 8–12×); eroding defensibility (a funded competitor or EHR incumbent adding AI); accidentally taking a big round that forces reinvestment-over-distributions.

Valuation math: ~$10M ARR at 80% margin, still growing at exit → ~$40–120M sale; at ~75% ownership = $30–90M to founder, plus millions/yr in distributions along the way.

2. The $10M ARR model (the cash-cow target)​

500 accounts at a blended **$1,500/mo all-in** ARPA. 80% margin ⇒ COGS ceiling ≈ $200/account/mo.

Revenue composition per account: subscription ~60%, the rest from payment-processing margin + marketplace/add-ons (which scale with client volume automatically). Example segment mix that hits $10M: a few done-for-you ($6k+/mo), some multi-provider ($2.5k), a bulk mid tier ($1.2k), and solo (~$0.9k all-in). Even the entry account nets ~$900 all-in — so $10M with ≤600 accounts is a premium play, not volume.

3. The $100M model (the platform path — the deferred option)​

$10M is a SaaS story; $100M is a platform/GMV story — you can't 10× the premium model (that'd need 5,500 done-for-you accounts). At ~13k businesses, subscription is under half of revenue; payments + marketplace GMV carry the majority ($3.6B GMV flowing through the platform ⇒ ~$50–60M from take-rate). This flips the levers: cheaper self-serve subscription to maximize logo count (every business is a GMV pipe you tax), then monetize the flow. Ordered levers to $100M: (1) more businesses, (2) more GMV/business, (3) take-rate — subscription price is least important at that scale. This is the Toast/Shopify model and is the hedge, not the current plan.

4. Pricing rationale​

  • Premium, not freemium. The old public page ($99/$499/$1,999, free-to-start, metered tokens) was the volume model and contradicts the cash-cow choice. Superseded by $999 or custom, charge up front.
  • Charge up front / emphasize annual. Mobile-subscription data: paying up front (even annual) beats free trials on LTV; personalized onboarding → excitement → willingness to pay. Hence annual-upfront with a first-year discount ($8,495) that renews at standard ($9,995).
  • Per-provider vs per-client: subscription is effectively per-account/seat (predictable, matches Healthie/Cerbo/Jane); per-client value is captured through processing % + marketplace/RCM % (which scale with active clients) — so no punitive per-client subscription. The active-client cap (1,000) both bounds cost and drives upsell.
  • Fewer options convert. Single public plan (or custom) beats a tier wall at the commit moment.

5. RCM / insurance economics (why 2–3%)​

  • Market: legacy full-service RCM = 5–8% of net collections (it's human labor — billers chasing denials/AR); AI-driven billers reset it to ~2.99%.
  • Our cost: Stedi is cents per transaction (claim $0.10–0.30, eligibility/status $0.08–0.30, ERA $0.08–0.20; no monthly minimum). A full claim lifecycle ≈ $0.35–1.10.
  • The arbitrage: legacy charges 6–8% for labor; Atlas automates the labor and Stedi is cents ⇒ we charge 2–3% tiered (below incumbents, still ~80%+ margin on the line). Bill on net collections, min $300/mo, payer fees at cost. Volume-tier the % down (higher billers expect lower %, and Stedi per-claim also drops → margin rises at scale).

6. Metered-AI reasoning​

The dividing line is text vs audio, not the four feature categories:

  • Text AI (copilot, doc/lab analysis, client chat) share the same per-token cost (Claude tokens) and differ only in volume ⇒ pool into one allowance shown with an example-consumption table.
  • Audio AI (Scribe per-min via Recall.ai, Call Center per-min via ElevenLabs+Twilio) have a fundamentally different, much higher cost structure ⇒ meter separately per-minute (folding a $0.90/min voice call into a token pool is misleading and a margin hole).
  • Overflow = auto-overage (AI keeps working, billed past the cap, alert ~80%) — no hard stop, since Atlas is the core interface. No "unlimited fair use" — there is always a cap then usage-based.

7. Margin model & COGS​

Target 80% gross margin/account ⇒ COGS ≤ ~$200 at $999. COGS stack: Atlas inference (Sonnet-default + prompt caching; Opus for hard tasks) + GCP infra + Terra (future, per connected client) + audio-AI vendor costs. Margin protectors: active-client cap + AI allowances/overage. Biggest risks to 80%: real GCP-per-active-client and per-minute vendor costs (verify against real bills). The +0.25% processing, RCM %, and add-on margins are revenue on top — they raise ARPA, they don't eat the margin.

8. Worked client example (reference)​

Annual + white-label, paid upfront: $8,495 platform + $2,995 setup + $4,788 (12×$399 white-label) = $16,278 due today (a client-specific round-down to ~$15,950 is fine). Renews ~$14,783 (setup one-time; white-label at $495/mo). Usage for a client at $60k insurance / $40k cash (assume 80% collected/processed): RCM $48k×2.75% + processing $32k×0.25% ≈ $1,400/mo → ~$16,800/yr additional. Year-1 total ≈ $33k.

9. Sequencing​

  1. Run the premium cash-cow to ~$10M ARR profitably, minimal dilution — pays the founder now, keeps control.
  2. At ~$10M, re-evaluate: if real GMV data validates the platform thesis and conviction in a big outcome appears, raise then (better terms, real proof). Until then the platform is a hedge, not the plan.