How Do You Buy Users Without Bleeding Money?
User acquisition only looks like gambling when you can't read the payback clock. Plenty of teams pour budget into installs and call it growth, then wonder why the bank account shrinks. The difference between a buy and a burn is one ratio, measured on the right window. Let's make it legible.
Q: What payback window is healthy?
For mini-games, a payback period under 90 days is the line most teams can stomach. Under 30 days and you can scale fearlessly; over 180 and you're betting the company on retention you haven't proven. The window isn't about being virtuous — it's about how fast you can reinvest yesterday's spend into tomorrow's installs. Short payback is what turns UA from gambling into a compounding machine.
Q: Do I look at CPI or LTV first?
LTV first, always. CPI is a price tag; LTV is the thing the price is buying. A twenty-cent CPI in a market where nobody pays is a worse deal than a two-dollar CPI where players stick and spend. Teams that optimize CPI in isolation are buying cheap installs that never return a cent. Anchor every decision on the LTV-to-CPI ratio, and measure LTV on the same cohort window you expect to pay back in.
Q: How do I total LTV with mixed monetization?
Add the streams, but don't double-count. Ads and in-app purchases overlap on the same player, so model per cohort: average ad revenue per user plus average IAP per user, minus the share who'd have paid anyway. A common rookie error is summing ad ARPU and IAP ARPU as if they were separate populations — they aren't, and that fantasy LTV is exactly what sinks the budget.
Q: How do I allocate budget without burning through it?
Cap spend per channel at a fraction of proven payback, then scale only the cohort that's already shown positive ROI. The instinct to "spend big to grow fast" kills more mini-games than bad games do. A runtime with a shared user pool helps here, because warm installs cost less than cold ones — your marginal dollar simply returns more.
Q: How do I spot fake traffic?
Watch retention shape, not install volume. Fake traffic installs and vanishes; real traffic retains a day-one that looks like your organic base. Also check completion of a trivial first action — bots skip it. If a channel's CPI is suspiciously low and its day-one retention is suspiciously flat at zero, you're renting ghosts.
| Window | What it means | Action |
|---|---|---|
| < 30 days | scale now | pour fuel |
| 30–90 days | healthy | scale with caps |
| 90–180 days | a stretch | only with proven retention |
| > 180 days | betting the company | pause, prove first |
