Lab · hands-on
F2P unit economics live
A lab with no code: turn retention, ARPDAU, CPI — and watch LTV jump nonlinearly and find where the game stops being profitable. The lesson's numbers, but under your hands.
How to use this
The key experience: drag retention r from 0.90 to 0.93 — and watch LTV jump by nearly +40% (it enters through 1/(1−r) and "explodes" near 1). The same relative bump to ARPDAU would buy you a couple of percent. Then push CPI above LTV — you'll get the red "unprofitable" verdict. A gacha calculator sits below.
The retention curve and LTV
Spending concentration: "90 / 9 / 1"
6%
dolphins 40%
whales 54%
share of revenue: ~90% of players (the free ones) bring ~6%, the ~9% "dolphins" bring ~40%, the ~1% "whales" bring ~54%. You cannot forecast from the average player — the tail decides.
Gacha: probability and pity
What to notice: 1) retention 0.90→0.93 → LTV +~43% (lifetime 10→14.3 days); the same +3% on ARPDAU → +3%, full stop. Retention compounds, ARPDAU does not. 2) at p=0.006 the expectation is ~167 pulls to an SSR — hence the pity systems that cut off the tail. 3) push CPI above LTV → the project burns money on every single install.
🏠 What's next
Go back to the lesson, section "🎮 Play / poke at it" — where to pull apart the monetization of a real F2P game with the eyes of a player and a tester (gacha rates in the store listing, battle pass, dark patterns).
Connections
from the lesson
F2P unit economics — the formulas behind the sliders: LTV = ARPDAU/(1−r), gacha 1−(1−p)ⁿ.
crossover
LLM NPCs — the cost per LLM turn multiplies by turns×DAU and lands straight in this equation; for an MMO it is a death sentence for cloud LLMs.
What to notice afterwards (observation checklist)
- Moved r and saw the nonlinear jump in LTV (retention compounding).
- Pushed CPI above LTV and got the red "unprofitable" verdict.
- Turned the gacha p and watched the expected number of pulls grow.