Click fraud isn’t new. Savvy marketers have known about it for some time. But the amount of fraud is so staggering – and the digital marketing industry’s response so lackadaisical – that the situation is approaching a tipping point that is sure to drive changes in media buying.
The Wall Street Journal picked up the topic in a recent article, reporting that “36% of all Web traffic is considered fake, the product of computers hijacked by viruses and programmed to visit sites.” This represents billions of dollars wasted every year on advertising seen only by bots and not the desired end users. This behavior is a byproduct of some perverse incentives by those who sell this media and poor oversight and analytics from the buyers.
At Acceleration Partners we’ve been very proactive and vocal about addressing and limiting fraud in online marketing. How? By taking the power away from bots and algorithms and empowering actual people. It’s not hard for companies to protect the integrity of their brands and prevent fraud—they just need their marketers to actually know where their dollars are spent and how their programs are running. They also need to ask questions when things don’t make sense and not be confused by complicated responses that don’t actually answer the question directly.
Know Where Your Ads Are Running
First and foremost, marketers need to understand where their ads are being placed. Lack of visibility opens brands up to fraud and also improper representation of the brand. The goal is to place ads on quality sites with content relevant to the brand. When marketers take the time to look at sites where their ads are running, this is usually obvious.
Part of this entails being vigilant about what ad networks and DSP’s you use. It is important to screen networks for trustworthiness and to avoid networks that place ads on hundreds or thousands of sites without proper visibility At the end of the day a human being may often need to be involved in making the call about whether a site is relevant enough to an ad and excluding those that aren’t and it takes sophisticated technology to monitor and detect fraud in what is an ongoing arm’s race.
CPA Is the Ideal Payment Model
One way to cut down on fraud is to pay for an actual sale or lead instead of a click. This cost-per-acquisition (CPA) model is ideal, but it won’t be possible to implement across the board. The next best thing is a cost-per-click (CPC) model followed by cost-per-impression (CPI). It’s important to note that there are no hard and fast rules. Fraud is still possible in a CPA model, it’s just harder to pull off than click and impression fraud as it requires a sale. However, each channel needs to be looked at under an ROI lens and at some point reviewed by human eyes. Also, no matter what type of media you are buying, your reporting should measure that media on a CPA basis, especially for e-commerce and ideally with multi-channel attribution to ensure that the spend is incremental.
Watch for Outliers
Make sure someone is watching for data that is outside of the norm, especially in display advertising. For example, if the click-through rate (CTR) from a site is well above the industry standard of 0.2% to 0.25%, that could very well be an indication of fraud. The same goes for huge spikes of traffic from a site without any apparent cause or relevance to the brand.
Looking at geography can also help. With both display and affiliate marketing, a good way to detect fraud is to look at traffic by country. Large amounts of traffic from Russia, India or Southeast Asia to a site that targets U.S. customers could also indicate fraud, especially if conversion rates or related metrics for that traffic are well below normal.
Understand Who Is Promoting Your Brand
In affiliate marketing especially, we believe it’s important to know who is promoting your brand. Develop a relationship with your best affiliates so that if something looks questionable, there’s a real person to ask about it and work with to address the problem. A person-to-person relationship can go a long way toward making sure promotion is consistent with the brand image.
In large part, the current fraud crisis is due to the fact that human beings have been removed from large chunks of the digital marketing process. Last year, for example, more than half of display ad buys were made using automated systems. By 2017 that number is expected to grow to 80%. If humans are carefully overseeing this automation, this can work, but too often the machines are left to do things on their own which has both brand and fraud implications. The marketing world now has plenty of big data, what we need is really smart people reviewing this data and using their judgment to limit fraud and ensure brand relevance in digital marketing. This way companies can expect stronger ROI on all their paid performance marketing programs.
Photo via JD Hancock on Flickr.