Blog · App Economics
💰 App Economics

App Economics 2026: The Complete Framework for Builders

📊 View 1-page infographic (share-ready PDF)

Before you ship an app, you should be able to answer: "what does each user cost me, what do they pay me, and how many users do I need to break even?" Most indie developers can't. This post is the framework that lets you.

The cost categories every app has

  1. Development cost — your time (or the people you pay), the tools you use, the AI subscriptions, the design work. This is mostly fixed up front.
  2. Variable per-user costs — what each active user costs you in: backend compute, database storage, AI API calls, third-party API calls (Stripe, push, SMS), bandwidth, support time.
  3. Fixed monthly costs — backend hosting floor, monitoring tools, your own AI subscriptions, domain, certificate, etc.
  4. Platform / payment fees — Apple's 15-30% App Store cut, Stripe's 2.9% + 30c per charge, etc.
  5. Support / customer success — the time you spend on user issues. Frequently underestimated.
  6. Marketing / acquisition — if you have any.

Revenue side

Revenue comes from one or more of:

AI API costs in detail

For AI-powered apps, this is the single biggest variable cost in 2026. Rough numbers, per million tokens (input + output combined for simplicity):

ModelInput $/M tokOutput $/M tok
Claude Haiku 4.5~$0.80~$4
Claude Sonnet 4.6~$3~$15
Claude Opus 4.7~$15~$75
GPT-5~$3~$15
GPT-5 mini~$0.40~$1.60
Gemini 2.5 Pro~$2.50~$10
Gemini 2.5 Flash~$0.30~$2.50

Prices fluctuate. Check provider pages for current rates. Prompt caching can reduce input costs 80-90% on repeat content.

How to estimate AI cost per user:

  1. Estimate AI calls per active session (e.g., 5 questions per session).
  2. Estimate average input + output tokens per call (e.g., 2,000 input + 500 output for a typical chat turn).
  3. Multiply: 5 calls × 2,500 tokens = 12,500 tokens per session.
  4. Estimate sessions per user per month (e.g., 4 active sessions = 50,000 tokens/month per user).
  5. At Sonnet pricing (mostly input-heavy): ~50,000 tokens × ~$5 / 1M tokens = ~$0.25 per user per month.

For your RDR2 Companion: at 4 free questions per user before paywall + occasional paid usage afterward, the AI cost per user is roughly $0.05-$0.50/month depending on engagement — well below the $2.99 unlock price, leaving room for everything else.

Backend costs at small / medium / large scale

Backend cost per user is non-linear — mostly fixed at small scale, mostly variable at large scale.

App-store and payment cuts

For indie iOS developers under $1M annual revenue, Apple's Small Business Program brings the cut from 30% to 15% — meaningfully changes the math.

Pricing models that work

A worked unit-economics example

Hypothetical AI iOS app: $9.99/year subscription, AI-powered chat with usage caps.

That's a ~35% contribution margin. Fine but tight. Levers to improve: use Haiku instead of Sonnet for routine queries (cuts AI cost 80%); add prompt caching (cuts cost on shared system prompts); raise price; add a tier above $9.99 with premium features.

Breakeven math

Fixed costs per month: backend baseline + your AI subscriptions + design tools + marketing.

Per-user contribution margin (revenue - variable cost): from the calculation above.

Breakeven users = fixed costs / contribution margin.

Example: $300/mo fixed costs, $0.29 per-user margin = ~1,035 paying users to break even. That's the number to anchor your growth plan around.

Mistakes that destroy margin


Platform-specific deep-dives: iOS, Android, Meta VR, Switch 2, Web, Windows, macOS, Apple TV, Apple Watch.

Sources & References
  1. Apple — App Store Small Business Program
  2. Anthropic — API pricing
  3. OpenAI — API pricing