As online ad fraud becomes increasingly prevalent and sophisticated, Dominic Mills says things are getting serious – and wonders whether there’s a solution for the industry.
I am coming round to the view that parts of the online advertising system resemble nothing so much as a communal cesspit gone toxic.
This cesspit is bedevilled by the rotting stench of fraud. The local residents all know there is a problem, but no-one has the responsibility to clean it up or drain it.
But don’t just take my word for it: read Marco Ricci’s impassioned opinion on Newsline last week about the growing prevalence, and sophistication, of the fraudsters and the way they are poisoning the online display market.
There are some cleaning agents that can help – Ricci’s employer, Adloox, Telemetry and others in the verification and brand safety space – but they can only do so if they are invited in by the ‘residents’ – in this case the publishers, the ad networks, the agency trading desks, and the demand- and supply-side platforms.
As far as I can see, and simplifying as only a technophobe can, you can divide the fraud into two broad categories: one, page fraud; and two, bot fraud.”
Let’s look at the problem. It’s nothing new to those who inhabit the Byzantine world of programmatic trading, but this story in May in the FT about the ‘bot’ fraud suffered by Mercedes, in which part of an online ad campaign was viewed more often by automated computer programmes than by human beings, brought home to the mainstream advertising community just how bad it is.
Ricci makes the point that one of the problems is that the fraudsters are invariably one step ahead of everyone else. Here’s a handy primer listing the various types of fraud, ranging from standard click farms, through to toolbar fraud, domain identity theft, ad stacking and several kinds of bots: phantom bots, ‘decepti’ bots (they mimic humans and are used in retargeting fraud), and ‘crypto’ bots (they mimic a smartphone running mobile apps).
Complicated, eh? And unless you have a PhD in computer science, utterly bewildering.
As far as I can see, and simplifying as only a technophobe can, you can divide the fraud into two broad categories: one, page fraud; and two, bot fraud.
Ad stacking, whereby multiple ads are placed on top of each other in one location, is an example of page fraud. Domain identity theft, whereby a fraudster hard codes a publisher domain into a legitimate ad unit, is another.
What about bot fraud? Bots effectively mimic the behaviour of people and are programmed to visit the fraudulent sites. Once crude (i.e. mindlessly repetitive clicking on ads at incredible speed), they are now increasingly sophisticated and more closely resemble human interactions. ‘Decepti’ bots, for example, can be programmed to show interest in a specific holiday or car, thus increasing their value to retargeters or advertisers buying a special interest.
And did you know that your average fraudster can actually rent a botnet for $600 or so down and $300 a month thereafter? No wonder it’s all over the place.
I wish there was a silver-bullet solution. But I can’t see it. It’s not as if anyone will be setting up an Interpol for ad fraud. Even if they did, the crims are always smarter than the cops.
I don’t understand why the clients don’t get more exercised. It’s their money.”
Theoretically, you could begin to eliminate it by preventing fraudulent inventory from entering the supply chain in the first place. But the providers – the publishers, the ad networks, the DSPs and so on – who make it available to the market appear to have little incentive. Why? Because they want as much inventory as possible – preferably cheap (and fraudulent inventory is invariably cheaper) – and don’t care where it comes from.
You could also tackle it from the buy-side, if the agency trading desks were more discriminating about what they buy. But again, money talks and they like to have lots of cheap crap to buy. The price of the total buy is averaged down, and no-one on the client side cares or understands enough to ask the hard questions. Result: the buy side is insufficiently incentivised to do anything.
It seems serious to me. Even if Marco Ricci’s estimate that fraud accounts for $11 billion of spend is an exaggeration, that’s still a lot more than chump change.
Me, I don’t understand why the clients don’t get more exercised. It’s their money.
MEC – go straight to the top of the class
I’ve fat-finger syndrome to thank for drawing my attention to the results of the IPA’s Excellence Diploma, something I wouldn’t ordinarily go out of my way to look for.
It was, however, a bit of a surprise. Of the 24 who passed, 14 were from media agencies.
The IPA describes the Excellence Diploma as an MBA in brand studies, and given the pedigree of the man behind the course, former BBH worldwide strategy director Nick Kendall, I wouldn’t demur.
And of those 14, no less than five – the single largest agency cohort – were from MEC.
Take a bow, MEC. You are indeed an example to the rest of the industry, and more particularly your media agency competitors.
For the record, the other media agencies whose staff passed were Carat; OMD/Manning Gottlieb; Zenith; PHD; Universal; and Initiative.
And what about some of those powerhouse, planning-centric agencies like, say, AMV, Saatchi and Saatchi¸ or Y&R, to pick a random sample? Nul, zero, none.
So what should we conclude from this? In no particular order, three things: 1) media agencies – or some of them – are deadly serious about getting a bigger share of the high-end, top-table, strategy action; 2) as a group, they are investing heavily in their staff; and 3) the traditional bastions of top-table strategy, the ad agencies, are either arrogant, complacent or don’t care enough about their staff.
This article was updated at 22:00 on Monday 21 July to state that ‘part’ of the Mercedes ad campaign ‘was viewed more often by automated computer programmes than by human beings’. The original incorrectly stated that ‘over half the impressions in a campaign were fake.’
