Over the past week I've been reading articles about how display is riddled with fraud and costing advertisers $6 million a month. From fake Twitter followers, to ghost publishers, this argument is going to persist for the web for a long time.
The argument of digital fraud in the advertising space is nothing new to search marketers.
Then: Click Fraud
Click fraud was once a very large problem in paid search. I remember a time when almost every RFP included questions about how our agency monitored click fraud. The number of times that question appears today in RFPs is down significantly, but with the resurgence of the argument I wanted to take a look into it again.
Google first introduced “invalid clicks” as a metric in 2006. This metric was intended to provide some insight into how much fraud was occurring and ease the minds of advertisers. Google identifies three areas they determine to be invalid clicks:
- Manual clicks intended to increase your advertising costs or to increase profits for website owners hosting your ads.
- Clicks by automated clicking tools, robots, or other deceptive software.
- Extraneous clicks that provide no value to the advertiser, such as the second click of a double-click.
Now: Top Sources of Invalid Clicks Today
I looked at my agency's clients to see what types of invalid clicks were coming through, and what areas they were coming from. The results were interesting:
- Desktops still have the highest percentage of invalid clicks (~6%). This is likely due to the lower amount of difficulty for hackers, and bots to be able to penetrate these devices.
- Mobile invalid clicks seem to be increasing (although within a tight range). My hunch is mobile invalid clicks come more from actual users with multiple clicks, or user error vs. actual purposefully fraudulent activity.
- Tablets had the lowest amount of invalid clicks, due to the higher degree of difficulty to hack these devices, as well as the improved ease of use (larger screens) vs. mobile.
Click fraud still occurs in paid search, and at some level always will. However, the engines at this point do a solid job at monitoring the problem. The issue for search marketers is to ensure the metrics don't go out of whack from a normal level of variance.
Meanwhile the focus will be on emerging media types where fraud is ahead of the development curve. Heck even 53 percent of Justin Bieber's Twitter followers are fake.