Bots Will Cost Advertisers $7.2 Billion. How-To Detect Fake Audiences
Advertisers are on pace to lose $7.2 billion globally to bots this year. Don’t be one of them.
Thanks largely in part to the speculation that bots are influencing the US presidential race on Twitter, automated social media bots are making headlines worldwide. While social bots aren’t new, there’s been a lot of discussion around them in recent weeks. What are bots? Does my audience have bots? And how do we find them?
Put simply: bots are algorithms that appear to look like “real people” on social networks. Some bots will follow you, some bots will share your content, and some bots will like what you post. Brands, politicians, and even Joe Schmoe’s use bot accounts to increase their follower bases and push or respond to messages.
For advertisers, the rise of bots is disastrous! Imagine discovering that 20% of your targeted audience are bots or the influencer you engaged in a multi-million dollar campaign has a fake following? Millions of ad dollars are spent promoting messages and content to audiences that will never become a customer. If only robots could buy your products or services, right?!
The challenge for advertisers today is that bots are getting smarter and harder to identify. They can respond to tweets, they usually have a steady following to follower ratio, and even have profile photos and individual bios. (It’s scary!)
So who you gonna call? Bot-busters!
^Did you get the reference?! At Affinio, we understand audiences through their connections. Bots are programmed to follow similar things, this makes it easy for the Affinio algorithm to detect them. So there’s only one thing to do… bust them!
While numerous bots exist, our algorithm excels at identifying two types of Twitter bots in particular.
The first I call “The Popularity Bot.” This bot is created to make an account appear to be more popular than it really is. (Fake followers.)
We can quickly spot “The Popularity” bots in a few ways, but here are the most common ways we accomplish this:
The first is by looking at a cluster’s influencers. Since our clustering algorithm is primarily based on “who you follow,” a group will immediately stand out as bots because they’ll typically have near-identical follow patterns that other groups don’t have. Within the Affinio dashboard, we make this easy to see: we have “Highlight Unique Influencers” buttons and filters that make these bot clusters stand out. In this scenario, the “Cluster Density” which shares how many of the cluster members know one another, would be extremely high (they all follow one another).
Another sign of a bot cluster is also an extremely low “Cluster Density.” Finding bots is all about finding the extremes or what is unnatural. Most audiences we analyze and the clusters within them have some friendliness between members that ties them together. When density is extremely low, it raises red flags and indicates an artificial community.
Low Density Bots
The second bot I call “The Conversationalist.” This bot is set up to shape a conversation around a topic to include something else.
We can spot “The Conversationalist” bots by analyzing the content that over-indexes in a cluster or clusters that have an extremely high number of tweets per month. Even though content used by cluster members isn’t itself a factor in determining how they group together, it’s still evident when one community uses a specific hashtag or shares a link (usually something spammy) extensively more than another cluster.
Our platform uses a trait-identification algorithm that helps make these over-indexing terms plain to see.
A Use Case for Bot-Busting! Identifying High-Value Influencer Audiences
One use case for bot-busting that is used by a variety of our customers is vetting influencer audiences. As shared by top PR firm, RF|Binder,
“Affinio has also been a valuable tool to identify spambots that follow influencers. Clearly, we wouldn’t want to work with an influencer who isn’t reaching actual people.”
Advertisers are continuing to be held to higher standards of measurement. If you are paying to reach a fake audience, while your vanity metrics may show success, you’re wasting your money.
To get real results, you need a real audience.
By understanding who exists within an audience, advertisers can ensure ad dollars aren’t wasted on targeting bots. They can also validate whether a partnership is getting exposure in front of the right audience.