Hydra, bots & click-farms: What marketers need to know

Two weeks ago, an ad-fraud scheme called Hydra was discovered by network security firm Protected Media to have drained $130 million in advertising spend worldwide.

This sophisticated operation evaded detection through simulating realistic mobile usage and allocating small amounts of traffic to a large number of ad networks.

Until Google and other big players find out a way to snuff out this fake traffic, budgets will continue to bleed. And after Hydra has been tackled, a new scam will rise, and the (Read more...) cycle continues.

The digital age was once hoped to be a marketer’s utopia with seemingly unlimited access to consumers across the globe.

But now, we know it to be a world plagued with bots, bogus accounts, fraud scams and click-farms — forcing marketers to take up the fight against these cyber monsters pilfering their budgets.

And even though the industry has reacted with refined fraud detection mechanisms, ad fraud has continued to evolve to dodge detection.

Because of this, utilizing accurate consumer data that filters out fraudulent identifiers has become vital for big and small companies alike to safeguard their advertising budgets against this threat and increase their return on advertising spend (ROAS).

The issue of click-farms

Like Hydra, the most successful fraud operations are the ones that simulate near-identical usage to real consumers, such as click-farms.

These operations usually consist of large groups of low-paid workers hired to click on paid advertising links. These workers can be operating up to 100 different phones at a time, each with its own fake account, email address, and user-profile.

Click-farms are often hired to improve website traffic, attract advertisers, and even used as a weapon to eat up a competitor’s advertising budget, and are especially prolific in countries that don’t enforce laws against click fraud, such as China, the Philippines, and Bangladesh.

The negative impact of these click-farms is two-fold: not only do they tank companies’ profits through pay-per-click fraud or fraudulent advertising that is never seen by a real consumer, but they also generate large amounts of erroneous data.

The vicious circle completes when this bad data finds its way into the marketplace where it is then bought and used by marketers to optimize their advertising campaigns and generate leads.

This is how this fraudulent activity deals double damage to companies’ marketing ROI. According to a report by Forrester Research, marketers spend 32% of their time fixing data quality issues and managing bad data.

To ensure that their advertising budgets are spent on targeting actual consumers rather than bots, here are some suggestions marketing teams can use for spotting fraudulent identifiers in their data sets:

IP addresses

Typically, 95% of all IP addresses have six or fewer emails associated with them. 95% of all emails have four or fewer IP addresses associated with them.

So, if an IP address is linked to hundreds or even thousands of emails, this is an indicator of potentially fraudulent activity. Removing these kinds of IPs from data sets can help marketers significantly increase the effectiveness of their campaigns.

Proxy server information

While the use of a proxy server is not a direct sign of fraud, it can be a useful indicator when analyzed with other variables to determine if someone is trying to hide their identity.

Consider using a proxy detection web service, which can detect the use of anonymous proxies and indicate if they are high risk. These services utilize public information and in-house resources to organize and maintain records on proxies in order to determine if they are highly used by fraudsters.

MAID characteristics

Click farms use a variety of practices to avoid being banned or spotted. However, marketers should try to avoid serving ads to accounts that exhibit typical click-farm characteristics.

This includes accounts that have been identified to like thousands of pages or other kinds of excessive traffic for extended periods of time.

By weeding out bad data from their data sets, companies can drastically improve their bottom lines. According to our findings, on average about 30% of consumer data is unusable.

Eliminating this bad data can improve ROAS by as much as 43%. Taken industry-wide, such savings would amount to billions of dollars in added revenues.

In a world where data has become our most valuable commodity, ensuring its accuracy must become our top priority.

David Finkelstein is a serial entrepreneur and founder of numerous internet companies dating back to the earliest days of the internet in 1994. He currently serves as the Co-Founder and CEO of BDEX, the first and the largest consumer data exchange platform in the U.S.