Agent-Skills.md

Agent Skills: Gaia Unit Economics Model

Provides the single shared, versioned unit-economics model for consumer one-off / in-app-purchase products, created in Discovery and re-run at every gate. Covers inputs (CPI, install->payer conversion, ARPPU, repeat purchase, ad ARPDAU), the gross-to-net stack (store commission, refunds, chargebacks, fraud, tax), LTV by segment, payback window, and a cashflow/working-capital check. Use whenever product economics must be initialized, updated with measured data, or evaluated at a gate.

UncategorizedID: frostaura/ai.toolkit.gaia/fa-unit-economics-model

Repository

frostaura/ai.toolkit.gaia
frostauraLicense: MIT
3026

Install this agent skill to your local

pnpm dlx add-skill https://github.com/frostaura/ai.toolkit.gaia/tree/HEAD/plugins/product/skills/fa-unit-economics-model

Skill Files

Browse the full folder contents for fa-unit-economics-model.

Download Skill

Loading file tree…

plugins/product/skills/fa-unit-economics-model/SKILL.md

Skill Metadata

Name
fa-unit-economics-model
Description
Provides the single shared, versioned unit-economics model for consumer one-off / in-app-purchase products, created in Discovery and re-run at every gate. Covers inputs (CPI, install->payer conversion, ARPPU, repeat purchase, ad ARPDAU), the gross-to-net stack (store commission, refunds, chargebacks, fraud, tax), LTV by segment, payback window, and a cashflow/working-capital check. Use whenever product economics must be initialized, updated with measured data, or evaluated at a gate.

Gaia Unit Economics Model

Scope and when to use

Use to maintain ONE canonical, versioned model that every stage reads and updates. No stage re-derives economics from scratch. Premium LTV is realized at install (finite, no tail); IAP LTV is back-loaded and whale-driven.

Use when:

  • Discovery must initialize model v0 (all inputs flagged as assumptions)
  • a specialist returns measured inputs (CPI, WTP, ARPPU, retention) to fold in
  • a gate needs the current net-of-fee LTV:CAC, payback, and cashflow read

Do not use when:

  • the product is B2B SaaS (different LTV/churn math)

Required inputs

  • acquisition: CPI per source/geo, organic share, blended CAC
  • conversion: install→payer rate, repeat-purchase rate, time-to-first-purchase
  • value: ARPPU (lifetime/per-cohort, NOT monthly), ad ARPDAU, segment mix
  • haircuts: store commission (15/30%), refund/chargeback %, fraud %, indirect tax %
  • timing: retention curve (D1/D7/D30+), paying lifetime, payout lag

Owned outputs

  • net blended LTV per install (IAP net-of-fee + ad net-of-rev-share)
  • LTV:CAC ratio and payback window
  • cashflow gap to break-even and required runway
  • a version stamp (v0..vN) and a changelog of which inputs became measured

Core workflow

  1. Initialize v0 with assumptions; label every input assumption-vs-measured.
  2. Compute gross revenue per payer and per install from conversion × ARPPU.
  3. Apply the gross-to-net stack: commission, refunds, chargebacks, fraud, tax.
  4. Add the ad component (eCPM × fill × sessions, net of network rev-share) for hybrid titles.
  5. Compute net LTV per install, LTV:CAC, and payback window by source.
  6. Run conservative/base/optimistic cases and sensitivity on install volume and conversion × ARPPU.
  7. Run the cashflow check: payout lag vs CAC timing; compute the cash gap to break-even.
  8. Re-version whenever a measured input replaces an assumption; keep the changelog.

Segments

  • non-payers (monetize via ads), minnows, dolphins, whales (top decile drives most IAP).
  • treat whale concentration as a fragility risk, not a static fact.

Anti-patterns

  • do not quote gross revenue at a gate
  • do not use monthly ARPPU or MRR framing (that is subscription math)
  • do not ignore the payout-lag cash gap
  • do not freeze assumptions as if measured

Handoff and downstream impact

  • give the money gate the net-of-fee LTV:CAC, payback, and cash gap
  • give specialists the current version and which inputs still need measuring

Completion checklist

  • every input is labeled assumption or measured
  • net-of-fee LTV:CAC, payback, and cash gap are current
  • the version and changelog are updated

References