Beat the Bots in Advertising
Published: January 16, 2015
Author: Jamie Gungler
I was on the edge of my seat with two browsers open and my pointer hovering over the button that would put me into the virtual waiting room. Foo Fighter tickets in Chicago were as good as mine. Flip of the clock, click of the mouse, and I was waiting …. waiting … frozen … sold out!!! I scrambled over to StubHub, and there were all of the tickets waiting to be sold at 2 and 3 times the original price. We humans didn’t even have a chance.
The culprit? The same things costing me and my display advertising clients money: bots.
Bots in Advertising
Similar to the programs that ruined my ticket purchasing experience, other, more malicious bots are costing me and my advertising clients money at work, too. According to IAB, at least one third of online traffic is non-human and in most circles it is believed to be closer to fifty percent. Why and how could this be? Because it’s in demand, that’s why. As IAB explains:
How bots work
Bots come in many forms, but generally, the malware is loaded onto an unsuspecting user’s computer, and that turns the computer into a bot. The bot connects with a botnet host and the host gives the bot directions. To make money in advertising, the bot starts surfing the internet to build relevant data pools so that the inventory looks appealing to an advertiser. The fraudulent inventory is offered through a dsp or audience extension platform and sold to the highest bidder.
The crazy thing is that, depending on the KPI, this may or may not offer the advertiser the exact inventory they are looking for. Certainly, if a campaign is paid out solely on impression levels, the campaign will simply meet the contractual terms – in this case, the inventory is cheap and performance isn’t an issue. A CPC campaign could experience the same contractual fulfillment because bots can be coded to perform any act a human can carry out. In this way, the KPI is still an issue as would be visits to the site … anything beyond an actual conversion could be simulated at a low rate, which creates more demand, and the cycle perpetuates.
How to detect and prevent bots
The Foo Fighters tried to beat the bots after the Chicago sales were dominated by scalpers. Their solution: they made box-office-only, in-person, pre-sale tickets available to the general public for their Los Angeles concert.
Here in the digital advertising world, we don’t have the ability to pull our digital placements offline, but we do have the ability to detect and prevent the fraud. Although bots are seemingly a staple on the internet, there are ways they can be exposed in your data and therefore avoided on different levels:
1. Monitor the trends for your inventory and question irregularities.
While a bot can be programmed to surf the web with human behavior, it is still a program and can only do as it is told. Over time, trends can be seen to pull out the fraudulent data and identify those sites that should be blocked.
If you’ve ever experienced a bot on your computer, you know that it brings things to a standstill. System performance can be assessed in real time and matched up to the big data to block the delivery of an ad before it hits an infected user.
2. Educate yourself and set standards within your organization.
3. Measure viewability (IAB suggests 70% in 2015 and a request for makegoods on anything over).
4. Ask your vendors:
-What % of your total traffic is proven human?
-What are you doing to prevent bot traffic?
-Are you implementing IAB standards?
5. Create whitelists from sites known to have valid converting inventory.
6. Work with certified vendors that meet either internal requirements or use a 3rd-party solution.
The best way to avoid the fraud is to be informed and become a part of the solution. Stay on top of the data and keep the conversation open.
And most importantly, know what to look for! Don’t get lazy in meeting deadlines and spend goals; know what you are getting into, and remember that if it sounds too good to be true, it probably is. Ads that aren’t seen or aren’t put in front of users who find value in the message are not worth your time. If your client wants to meet only spend and CPA goals, they may not be the client for you.