Fraud, Invisible Ads and Silly Targeting

Robots Watching TV
(Photo: “Robots Watching Telly” from Nice Paper Toys.)

Mercedes-Benz recently accused Rocket Fuel, the giant ad network, of fraud, asserting 57% of the impressions the car-maker bought on the network’s websites were seen only by non-human software bots. From the FT:

Part of a recent Mercedes-Benz online advertising campaign was viewed more often by automated computer programmes than by human beings, according to documents seen by the Financial Times.

The ads were inadvertently placed on to fraudulent websites by Rocket Fuel, a Nasdaq-listed ad technology company that went public last September with a market capitalisation of nearly $1bn.

Rocket Fuel, in a blog post, refutes those numbers, claiming only 6% of the Mercedes-Benz impressions were fraudulent, and that they replaced them with ad impressions served to actual humans “before any bills were even sent.” In the same post, they tout their prowess in identifying and declining fraudulent inventory. “We reject approximately 40% of all ad space daily due to its failure to pass our own bot and brand-safety screens.” Good for you, Rocket Fuel, but bad for the rest of the industry. Forty-percent?? What’s 40% of $43 billion in digital ad spending?

Of course, the industry recently made a small leap forward. At least regarding the online ads that are aimed at humans — 54% of which, according to Comscore, are delivered to parts of the web that human eyes can’t see, below the fold or otherwise outside the browser’s viewport. The IAB has formalized a new viewability standard. No longer will publishers or ad networks (those that comply, anyway) charge for ad impressions delivered outside the visible, on-screen space on a consumer’s screen. From now on, half (or more) of the ad unit must have the opportunity to be seen, and must stay there for at least one second. If you’re starting from zero seconds and invisible, I guess, this is progress. Modest, incremental progress, but it’s a start.

Numbers like that — 40% is fraud, 54% can’t be seen by humans — can depress a person. But the ads that aren’t invisible or software scams, at least they’re super targeted and awesome, right? It’s downright scary how those online ad algorithms know absolutely everything about us, right? Sometime I’m not so sure.

Roman Mars Tweet

Earlier this week I saw this tweet from Roman Mars, the prominent design and architecture journalist. The New York Times design columnist Allison Arieff calls him “the Ira Glass of design.” Meanwhile the marketing team at School of Visual Arts wants to send him back to school, and is spending marketing dollars to pursue its case.

It struck me that those emails and postcards would qualify as excellent, near-perfect targeting in the world of digital advertising. Last weekend I did two things online. One, I visited the website for a touristy Chinatown restaurant (Z and Y) to get their address, and, later that night, ate there with my in-laws. Two, I bought a Welsh Love Spoon as a present for my daughter. For the rest of the week Google served me ads for Z and Y Restaurant and promotions for Welsh Love Spoons.

Silly, right?

Do the ad-targeting algorithms think I’m due to go back to Z and Y so soon? And if I did go back to Z and Y a week later, shouldn’t the chefs and waiters ad Z and Y, not Google, get the credit for my return trip? And the Welsh Love Spoons. Is there something in my personal Big Data that says I’ve suddenly become an obsessive Welsh Love Spoon collector? I can attest to the fact that these ads connected with human eyes — for a full handful of seconds! — but I’m still tempted to call them a scam.

  1. # Tac Anderson said: June 9th, 2014 at 8:36 am

    The problem is that there is no closed loop. Unless you bought the spoons using Google Checkout, they have no way of knowing that you bought them, and they have no way of knowing that you ate at Z and Y, so from their perspective you just showed an initial interest.

    Now, Google could probably get a much better approximation. You probably have a gmail app and Google maps on your phone. Gmail could look for receipts from your online purchases and maps could check your location against that of the restaurant. But the question is, would they (once they overcame the technical hurdles of plugging all those pieces together). They would be limiting even more add opportunities, which would be good in the long run, but bad in the short run.

    Or maybe you all of a sudden have developed OCD and only eats Chinese food with wooden spoons.

  2. # Chas said: June 9th, 2014 at 8:52 am

    The OCD isn’t a sudden development, but my variety of the disorder doesn’t yet extend to Chinese food or wooden spoons!

    What you are saying is true — since Google doesn’t know if I’ve made the transaction or not, my web behavior might indicate unfulfilled demand. But it’s a funny thing that Google gets a targeting premium from the advertiser when a consumer expresses demand that is unrelated to services provided by Google. The car salesperson who hands me the keys when I walk onto the lot asking for a test-drive doesn’t make much money; why should s/he? S/he didn’t hunt me down and convince me to consider that particular brand of car, s/he just harvested demand that I built up through some other means. In this case I learned about Z and Y by texting a friend who knows Chinatown restaurants, and about Welsh Love Spoons on Wikipedia helping my daughter with a school project. Why does Google get the credit (and the money) for my commercial intent, when I came upon that intent from others?

  3. # Tac Anderson said: June 9th, 2014 at 9:07 am

    Because unlike the car salesman, Google has figured out how to monetize intent, regardless of where it originated, and advertisers are desperate enough to pay them for it, and lazy enough to live with the faults in the system (for now).

  4. # Chas said: June 9th, 2014 at 12:39 pm

    Good point. Why should I blame Google for collecting money from lazy advertisers?

Leave a Reply