Spent $220 on Google App Campaigns — 60% of Installs Were Bots

A solo developer's real-world experience has ripped open a long-ignored black hole in mobile advertising: the "users" you pay real money to acquire may largely be bots. This article — which scored 274 points and sparked 154 comments on Hacker News — comes from a game developer's firsthand test of Google App Campaigns. The results were unsettling: roughly 60% of the installs from a $220 spend came from non-real users.

The Experiment: $220 Buys a Hollow Victory

The developer ran Google App Campaigns for their game with a $220 budget. On the surface, the dashboard looked great — installs were rolling in, and the conversion numbers were tempting. But when they dug into the behavior of these new "users," the problem became obvious: nearly all of the newly installed devices showed zero real interaction. Retention curves dropped to zero almost instantly, and behavioral patterns were unnervingly uniform and mechanical — classic hallmarks of automated scripts, or bot farms.

In other words, more than half of the "install conversions" reported by the ad platform weren't real players at all — they were mass-controlled devices or emulators. For an indie game that lives and dies by in-app purchases and active engagement, this means the bulk of the ad budget was completely wasted. Worse, it may have polluted app store ranking signals and corrupted future user profiling.

How Bot Farms Work

At their core, bot farms use large numbers of real or virtual devices to execute "click ad → install app" actions in bulk, gaming ad network attribution to claim conversion payouts. Under cost-per-install (CPI) or cost-per-action pricing models, every "install" translates to a charge for the advertiser — and that money flows straight to whoever manufactured the fake installs.

The technical barrier for this kind of fraud isn't particularly high. Device farms, emulator clusters, and hijacked apps with embedded SDKs can all serve as sources of fraudulent traffic. What makes it especially tricky is that bots deliberately mimic normal install behavior, making it hard for attribution systems to flag them in real time. By the time a developer notices the anomaly in retention and monetization data, the money is already gone.

The Gray Zone of Platform Responsibility

What really struck a nerve in this article was its challenge to ad platform accountability. As a giant with massive data and sophisticated fraud detection capabilities, Google is theoretically far better positioned than any individual developer to identify and filter bot traffic. Yet in practice, advertisers are left to discover the fraud themselves, gather their own evidence, and navigate an opaque and unfriendly refund process.

Judging by the Hacker News discussion, this is far from an isolated incident. Many developers report running into similar fake install problems across various ad networks — not just Google. Solo developers with small budgets and no dedicated anti-fraud teams are consistently the easiest targets. Since platforms are both the rule-setters and the ones collecting payment, they have little inherent incentive to proactively clean up "traffic" that generates revenue for them.

How Independent Developers Can Protect Themselves

For budget-constrained developers, this case offers a few practical takeaways:

• Don't just track install volume — track behavioral depth. Day-one retention, key action completion rates, and monetization conversion are the real indicators of traffic quality. Pretty conversion numbers paired with zero retention are almost a guaranteed sign of fraudulent traffic.
• Break down data by channel and region. Bot traffic tends to cluster around specific device models, OS versions, or low-cost regions. Cross-referencing these dimensions can quickly surface anomalies.
• Use anti-fraud and attribution tools. Third-party mobile measurement partners (MMPs) typically have built-in fraud detection mechanisms and are more reliable than trusting ad platform self-reported data alone.
• Document everything and file disputes. The moment you spot anomalies, screenshot and export your data, then request a refund from the platform. The process is painful, but it's one of the few ways to recover losses at this point.

A Final Thought: The Trust Crisis in Digital Advertising

This $220 experiment reflects a deeper crack running through the entire digital advertising ecosystem. When the metrics used to measure success — installs, conversions — can be cheaply fabricated, and when the party doing the billing has no real incentive to pop the bubble, advertisers — especially small developers with no bargaining power — end up as the ones footing the bill.

For developers, the takeaway is clear: rather than blindly trusting the polished numbers in a platform dashboard, build your own framework for evaluating "real users." And for the industry as a whole, this case is yet another reminder that ad fraud isn't a fringe issue — it's a systemic challenge that demands honest engagement from platforms, regulators, and developers alike.