Protecting an ad account from invalid traffic

The risk is different for publishers

A media buyer who ignores bot traffic loses money on that campaign. A publisher who ignores it can lose the ad account that funds the whole site, which is a different order of problem and deserves a different screen.

Google states plainly that invalid activity can lead to account action, and it is worth reading their description in their own words rather than a summary of it.

Watch SIVT, not total bots

Search crawlers and monitoring tools hit every site on the internet and exchanges filter them on their own side. What draws attention is sophisticated invalid traffic: automation behaving like a person, and clients impersonating a known crawler. Watch that number and largely ignore the other one.

The trend arrives before the problem

By the time a share is obviously bad, the review has usually already started somewhere you cannot see. A site sitting at a comfortable level whose SIVT share tripled in a week has something happening now, which is why the risk reading warns on the direction rather than only the level.

The usual causes, in the order worth checking

1. Bought traffic. If any part of your audience is purchased, check that source first. It is the most common explanation by a wide margin.
2. A single referrer sending a burst. Look at sources sorted by bot volume rather than by total.
3. Your own testing. Uptime monitors, load tests, and preview tools all generate traffic that looks automated because it is.
4. Someone else pointing traffic at you. It happens, it is not your fault, and it is still your account.

What no tool can tell you