Sean Penn went into the lion’s den to discover the method of a sophisticated Drug Lord preying upon society’s huge economic, ever-present underbelly that nobody like to discuss.
Penn is controversial, but his most endearing quality is that he lays it on the line. From bad guys in Mystic River to a gay activist in Harvey Milk, to the real world of disabled people in I Am Sam…his roles and filmography tell the truth regardless if you want to hear it.
The same is true in Programmatic digital advertising. Much has been written, but like El Chapo, just because you take down a kingpin…it still continues.
In the U.S., programmatic display is now a $20 billion industry, and the majority of all display buys (eMarketer, 2015). Where there’s money like that: bad guys are there too even though we may not see them.
So who are the Drug Lords in Programmatic? It’s not as much about the who as it is the what and the where eating at the $20 billion. Here they are:
1. Midnight to 6am
What happens at midnight?
The clock for publishers and Exchanges gets reset. And every day publishers and exchanges need to make money.
When does fraud occur? Midnight to 6am.
What changes? Effectively they say: I need to make money and I’ll take anything I can get. Midnight to 6am becomes the fraudsters playground.
But data you get as an advertiser or agency is by day (not by hour). Impressions, clicks, click-through-rate, cost, cost-per-click, cost per impression, etc. No hourly data comes your way. Thus with most systems, as an agency or brand…you’d never know what portion of your ad buy is being delivered at the most inopportune time. Not a clue.
But with a measurement platform that sees hour by hour the distribution of impressions and you may be shocked. Kind of like this:
Take your mouse over these peaks (showing day, date, and impression count for the hour) and look for peaks with an hour of “0:00”. This is 12 midnight. If you see strange high peaks of impressions, at 12 midnight through 6am, that is daypart fraud. A huge portion of display fraud happens between 12 midnight and 6am in the morning, because publisher sites have a new day at midnight and need to serve ads (any ads).
90% of a customer’s ads ran between midnight and 3am. After repeated conversations with the network, the customer called the network and fired them.
Midnight-6am is one of Programmatic’s Drug Lords. There’s crime in New York City, and this time-frame is when Drug Lords are operate. Both of these facts won’t change, but you can use platforms to detect, monitor, and shut it down.
2. Video (Except This Video)
Your trusted inventory: YouTube, Facebook autoplay, twitter autoplay. This is the good stuff. That being said, it’s not 100%. Google got it’s ass kicked by fraud in 2003, made huge investments to combat it (sacrificing revenue). It still came back to haunt Google with AdSense which was a fraudulent playground until 2014…but it cleaned up AdSense and Google’s DNA has the best in class anti-fraud control (though no perfect) unmatched anywhere.
But with just about every other source of video inventory apart from the above, there’s more fraud in video than in any other sector of the industry (why: higher CPMs). If you’re a programmatic Drug Lord, video is your new heroin.
Lots of small and medium-sized ad tech companies as sellers of video inventory are under pressure for earnings & revenue. The conflict is: turn a blind eye to video fraud that will happen just about everywhere except YouTube, Facebook, and twitter (it’s going to happen)…or make revenue numbers. Or, spend money (a lot of money) to combat video fraud and make your revenue numbers even worse. And thus the conflict of interest in video: turn a blind eye and make revenue, or do something which is like fighting the cartel which will keep on operating (except in well-policed neighborhoods).
3. Viewability (vs. the Cookie)
Cookies taste good. Cookies are addictive. Cookies aren’t going away.
They serve a needed purpose in programmatic, but cookies are like dessert (it’s a diet imbalance).
I love ice cream, I love cookies…but there’s a place for it. To have every meal consist 100% of dessert isn’t going to do much for your health in the long run.
As it is with cookies and viewability.
In programmatic display, 51% of ads are not viewable according to C3 Metrics.
So what’s the big deal?
a) Imagine if you’re Procter & Gamble and you order the raw materials to make diapers. Now imagine 10 trucks were supposed to show up with the raw materials, and only three trucks showed up? There would be hell to pay everywhere. Diaper orders wouldn’t get fulfilled, contracts likely lost.
b) But the really big deal is that the entire ecosystem of programmatic display optimizes on cookies…and if 51% of the source data is based on ads that are not viewable, the ecosystem continues to optimize against what? Against non-viewable ads. Automated/Artificial Intelligence (e.g. Programmatic) will optimize against dessert. It tastes good right now (I know it’s not good for me as my sole diet) but what the hell, let’s have even more. Ice cream and cookies for breakfast, lunch, and dinner 24 hours a day.
Sure there’s some progress on this. But predicted viewability is very different than actual viewability.
There are three components to viewability: consumer behavior, page uniqueness, and exchange infrastructure.
Anyone guaranteeing viewability can’t actually guarantee viewability without make-goods because of these three components.
No Programmatic source can impact this. If Mrs. Smith clicks on a page, the display banner ad call (server request) is made to serve the ad, and Mrs. Smith clicks away to another link on that site before the ad is either loaded or is above the fold…no programmatic vendor can make Mrs. Smith go back, let the ad load, and then scroll down to have the ad be viewable. Mrs. Smith may be on a fast connection, or Mrs. Smith may be on a slower connection or a 4G connection with a ton of clogged activity on that connection, but it’s her unique connection and Mrs. Smith is free to surf how, when, where, and on what device she wants…which affects actual viewability.
Each page is unique. It’s design, it’s content in terms of copy, it’s content in terms of graphics, and it’s speed of page load. Thus either the position or speed (including cache which can be unique to each device) of a page affects how a display ad loads in terms of speed and in terms of where in the browser viewport it loads, and will be different page by page which affects actual viewability.
Here’s the secret nobody will ever tell you. 15-20% of all display ads never even load. Right off the bat, 15-20% waste. The reason: combination of the above two plus Exchange Infrastructure. An “impression” is simply a request to serve an ad. An “impression” does not mean that the ad (won at programmatic auction) is actually delivered on the consumer’s page. It’s a supply side metric, not a consumer side metric. Apart from the first two items cited above, the efficiency and speed of the Exchange’s Infrastructure (everywhere in the country) can also affect viewability. If an exchange is using older technology (’72 Buick) versus a Tesla to deliver that ad, speed counts. If the vehicle/infrastructure doesn’t get there in time to load on the page (which is not in isolation to consumer behavior and page uniqueness) it will also affect actual viewability.
It’s a complicated picture, but there is one simple truism: no matter how efficient advertising is, it won’t work unless it’s actually displayed. Cookies don’t equal viewability and if your diet consists of cookies and ice cream every meal, every day, then your body is not recieving the essential nutrients you need to survive. It is the third Drug Lord of Programmatic.
Think your day job has nothing to do with El Chapo? Think again. Protect your own.