A difficult problem for advertisers to come to terms with is the size and scope of online ad fraud.
Everyone knows it's a headache, but no one knows what kind of headache. Is it just eye strain or am I having a stroke?
If you don't already have a headache, trying to reconcile the various estimates of the size of the problem will give you one.
- Group M, a WPP-owned media company, says that ad fraud is "significantly contained" and that "2% of impressions purchased by the biggest advertisers in Western markets remain non-human."
- The&Partnership, another WPP-owned company, says that ad fraud constituted about 20% of online spending or about $12.5b last year. I think it's cute when two divisions of the same corporation disagree by 1,000%.
Association of National Advertisers claims that about $7.2 billion was stolen last year by online ad fraudsters, which would amount to about 11% of total ad spend.
- The World Federation of Advertisers say that fraud could be as high $30b which would be almost 50% of online ad spend.
- Last year Facebook announced
that they had cancelled a buying platform they were testing when they found that 75% of the inventory coming into the platform was "valueless."
In other words, everyone has an opinion and no one has a clue.
Most of the people I talk to who appear knowledgeable guess that the number is probably between 20% and 40%. But the problem is, no one really knows.
From what I understand the difficulty in ascertaining a real number is that the people who measure fraud (the cyber-security crowd) all have different criteria for deciding whether a visitor is a human or a bot (WARNING: DUMBASS BLOGGER EXPLANATION COMING UP...)
Remember, in the digital realm everything is either a 1 or a 0. There ain't that much difference between a human 1 and a robotic 1. So the security boys have to set up filters that look for patterns
and try to interpret what's a human pattern and what's a robotic pattern.
Meanwhile, the bad guys are constantly sending feelers to see which patterns they can invent will pass through the filters. Once they find a winning pattern they bombard the system with fraudulent traffic and clicks.
Consequently, when a cyber-security firm reports that 5% of traffic is fraudulent, what they are really saying is that their nets were able to catch 5% of the fish, but they don't really know how many fish got through the nets.
In one instance, as a test, an
ad fraud fighter sent 100% robotic traffic through a network and only 17% of it was identified as such by a leading cyber security firm.
Fraud is a major problem. In an environment in which over half of display ads aren't even visible and adtech middlemen are scraping 60-70% of working media dollars, rampant fraud is the last thing we need.