The pattern that gives it away
Here's the pattern that gives it away. Your install count looks healthy, maybe even cheap, and on the dashboard it feels like things are working. Then you look one step deeper: how many of those installs opened the app, started a trial, made a purchase. And the line is flat. Thousands of installs, almost no behavior. That gap, installs without humans behind them, is the signature of fake or junk traffic.
It usually traces back to the source. When you optimize purely for installs and chase the cheapest possible cost, you create a perfect target for low-quality networks and incentivized traffic: places that can deliver a flood of "installs" that were never going to use your app. Sometimes it's outright bots. Sometimes it's real people paid to install and immediately delete. Either way, you bought a number, not a user.
The fix: move the goalpost past the install
The most reliable fix is to stop buying the number. Instead of optimizing for the install, optimize for an in-app event that a real user has to do: open the app, complete signup, start a trial. Bots and incentivized traffic are good at producing installs and bad at producing genuine usage at scale, so when you move the goalpost past the install, the junk has nothing to fake and the platform stops chasing it. This is the same logic as optimizing a subscription app toward paying users instead of trial signups: aim at the behavior you actually want, and the quality follows.
One simple early-warning sign worth internalizing: a cost per install that looks too cheap to be true usually is. We've run installs legitimately under $1, so cheap isn't automatically fake, but cheap plus no downstream behavior is the combination to distrust. The price is only good news if the user behind it is real.
So before you blame the creative or the targeting, check the gap between installs and actions. If installs are pouring in and usage isn't, you don't need a better ad. You need to cut the source and change what you're optimizing for.