If you have used the Google Display Network to run display campaigns for years, you know that it is easy to notice various patterns. It is also noticeable that some of these patterns seem to work better than others.

For example, we know that Google can determine the gender a person identifies with. One of my favorite patterns to use involves this principle, and it is: only target customers whose gender Google is able to identify. 

The reason for this may not be what you may think, as there isn’t one particular gender that tends to work better than others across all ad campaigns. Even though people might think that knowing a user’s gender can help predict purchasing and activity patterns, this is not necessarily true. Saying, for example, “people who identify as female tend to buy more than people who identify as male,” is not valid. It is true that this principle may apply to particular niches or categories. But if someone thinks that’s true overall, I have some data to show you and a bridge to sell you, while we’re at it. 

But, there is another case in which knowing a user’s gender is important, and it has nothing to do with what gender someone may or may not identify with. The case to look out for is not what the person’s gender is, but when Google can’t figure out how a user identifies. Very consistently, this unidentifiable user seems to bounce, never convert, and in fact, it seems like the user is not human at all. 

Why is it important to look out for this? Because the main reason Google can’t figure out how a user identifies is probably because it is not even human!

Let me explain. 

Google keeps track of a user’s online behavior across all Google platforms and properties. Based on this behavior, Google is able to predict the gender that the user identifies as. And Google is very, very good at doing this. 

In fact, Google is so good at predicting this information that if it can’t quite figure it out, we can consider it a red flag. It’s most likely because the user isn’t human!

Spammers program their bots to spam, not particularly to “spam as a self-identifying girl would,” for example. For this reason, bots rarely show any self-identifying gender behavior as they browse online. 

As a result of this, we know that when Google can’t seem to identify a user’s gender, that said “person” is more likely a bot. Say, from a fifty percent chance on the Google Display Network—without our tool, of course—to a ninety percent chance, to use a rough estimate I often use as a rule of thumb. 

So, an easy way to reduce the probability of fraudsters sending bots to your site is to exclude anyone whose gender Google can’t identify from your display campaigns.