Feed the Platform Real Purchases, and Bot Traffic Drops on Its Own.
A thin, imperfect purchase signal did more to stop bot traffic than any targeting change did, because bots are a symptom of a missing data feed, not a targeting failure.
The instinct is to fix targeting. The actual lever is data.
When bot traffic spikes inside an ad account, the reflex is to tighten audience settings, add exclusions, or pull back a traffic objective's exposure. None of that addresses why the bots showed up in the first place. A campaign optimizing purely for clicks or landing-page views has no real business outcome to chase, so the algorithm optimizes toward whatever generates activity cheaply, and low-quality or automated traffic often clears that bar more easily than a genuine buyer does.
A thin signal is still a real signal
In one documented account, sending back even an imperfect purchase signal, an Event Match Quality score around 5 out of 10, far from a clean implementation, changed delivery. Bot-driven traffic dropped once the platform had something real to optimize toward, even a partial one. Campaigns still running on a pure traffic objective, with no conversion signal feeding back at all, kept attracting the same low-quality volume.
This is a data-infrastructure problem wearing a targeting costume
The platform isn't malfunctioning when it fills a traffic objective with junk clicks. It's doing exactly what it was told to optimize for, in the absence of anything better to chase. The fix isn't a smarter exclusion list. It's a reliable purchase signal flowing back from your own systems, verified against what actually happened in the store rather than modeled from a partial pixel fire. Once the platform has a real outcome to target, low-quality traffic becomes a worse trade for its own optimization, and it moves away on its own.