Invalid traffic (IVT) has been a subject of deep concern for marketers and media companies alike for several years as plenty of news headlines can attest. But despite universal agreement in the industry that it’s a significant issue, there are still common misunderstandings throughout the digital media industry about how IVT is generated, how to find it, and even what it is.
We recently hosted a live webinar, “Shedding a Light on Invalid Traffic,” covering those topics and more. Dan Fichter, the VP of Engineering at Oracle Data Cloud who directed Moat’s IVT detection technology for more than five years, walked through the basics of IVT to define it, show why and how it’s done, and what the industry can do about it. Callie Reynolds joined him to provide her perspective as head of account marketing for Moat customers.
This blog post captures only a few of the key takeaways covered during the session. Please view the webinar here to learn more. Tackling the IVT problem can only happen once the industry has a clear understanding of the challenge it faces.
Invalid Traffic isn’t always malicious. In our industry, IVT is often conflated with ad fraud—a closely related issue. Though there’s evidence the industry is making progress combatting it, ad fraud remains a huge problem. The Association of National Advertisers (ANA) estimates the dollars lost to ad fraud dropped 10% in 2017 from the year before, but that still amounts to $6.5B in lost ad spend.
However, by definition, fraud is a deliberate crime, while much of IVT is harmless. There are plenty of legitimate reasons for a bot to generate an impression. For example, we depend on spiders from search engine giants like Google to make the web practical.
Of course, that doesn’t mean a marketer pays for impressions delivered to a bot. When we talk about IVT, we simply mean these impressions shouldn’t be paid for because they weren’t delivered to a person (or, because they’re delivered in the wrong way, per the next webinar takeaway).
IVT can happen even when a person is visiting a site. Some ads technically may be delivered to a human’s screen, but remain totally hidden from the user. In these cases, those impressions should not count against anyone’s spend.
Consider these examples of hidden ad types.
Stuffed ad: One ad may be in-view, but the iFrame contains more ads outside its viewport, causing them to be hidden.
One-by-one ad: The ad is in an iFrame smaller than the creative that doesn't expand for the entirety of the user's session.
Invisible Ad: The ad is transparent for the entirety of the user's session.
When it comes to IVT rates and measurement, the denominator matters. Discrepancies between measurement vendors are largely attributed to their different measurement footprints. For instance, Moat benchmarks between 2% and 6% disprove there are a large number of channels with IVT rates as high as 90%.
This chart shows how most video channels have an IVT rate lower than 10%, as measured by Moat, but there are plenty channels with more.
IVT is a hard problem. When it comes to digital advertising, a number of factors make invalid traffic a particularly difficult challenge for the industry.
Perpetrating ad fraud can be easy. Legitimate tools and services like automated browsers and the public cloud are widely available and can be used to commit fraud. During the webinar, Dan Fichter presents an example illustrating how simple it is to create false demand for ads.
More and different types of data improves IVT detection. Access to different types of resources and information—for instance, what percentage of a site’s audience can be tied to an offline purchase—helps us ask new questions to determine true IVT levels.
We all have a stake in combatting invalid traffic. Invalid traffic is changing, and the problem will get better only after we untie as an ecosystem and end the financial incentive for ad fraud.