Wait, who sees my credit card bill, again?
We've done a lot of work here at Adweek on ROI data, and a few readers have asked that we explain what on earth we're talking about, because it's a somewhat scary phenomenon to consumers—especially in an age when surveillance is such a hot topic—and a thrilling opportunity to advertisers.
Here's the short version: Everyone in advertising is buying exhaustive records of your purchases—all your purchases—and comparing them to your viewing habits so that they know which ads you saw and whether or not they changed your behavior.
If you feel like this is kind of invasive, that probably means you understand me so far.
All of Your Financial Information Is for Sale. Here's How It's Collected.
- When you shop frequently at a store, you get a points card so that you can get a discount or coupons. Stores give these away like they're going out of style, ostensibly to reward loyalty—Kmart, Walmart, Target, Walgreens and CVS all do this. Hell, I have a Godiva free-truffle-of-the-month card (SO WORTH IT). So that takes care of consumer packaged goods, or CPGs, which are a class of merchandise worth about $2 trillion, where growth has slowed. The category is defined a little more specifically as "products that need to be replaced frequently"—potato chips and toothpaste and toilet paper, as opposed to washing machines and armchairs.
- Two main companies, Acxiom and Experian, collect this data, among other data sets. Do you remember a few years ago when you ordered a credit check to see if you could qualify for financing on that pirate ship you wanted to buy? You probably got it from one of those guys. Now that you have your pirate ship, you've forgotten all about them, but they are expert data miners.
Car registration data is acquired from companies that have access to DMV records, which became searchable by pretty much anyone who wants it thanks to a couple of clauses in the Driver's Privacy Protection Act of 1994. Jim Moran wrote that bill to keep people who work at women's health clinics from being murdered by enterprising activists who were using the names of abortion providers to find their DMV records, which included their addresses. In his defense, 1994 was too long ago to foresee the use of anonymized data at scale by companies like Polk.
- The DPPA in particular is why companies like Acxiom, Experian, and other data brokers (and the companies that use data brokers) are skittish about publicity—the clauses that allow the use of this data are designed to expressly forbid the direct identification of anybody involved.
- So data brokers tip-toe right up to the edge of that line—they whitelist everything they have, meaning that they strip out names and addresses except for the ZIP +4 code (you know, 55555-5555, instead of just 55555. It designates a particular location—a city block, or one really big high-rise, or maybe a business that gets a ton of mail every day).
Now, the problem with matching auto data to ZIP+4 is that ZIP+4 is actually not that easy to whitelist when it comes to, say, a Maserati on the Upper East Side. Chances are only one guy on that block owns a Maserati. A source told me a year or two ago that what data brokers are afraid of is "backing into identity," and that's a good way to put it. You're not setting out to invade privacy, you just want statistics so you can plan better. But you can get there by accident pretty easily, and that's why these companies are so careful.
So we have consumer packaged goods, we have cars and we have houses in more or less the same way—but what about pharmaceuticals? They don't know I'm on Zoloft, do they? Actually, you buy your Zoloft from CVS, so yes, they do. Mortgage payments? Fruit? The couch you bought at Macy's?
- Well, you probably bought all that stuff with a credit card, and there's a company called Argus that one data company executive told me provides "the majority of credit card transactions" to them for the same purposes. Argus is good at that, too—Breitbart and Judicial Watch identified it among contractors the Consumer Financial Protection Bureau is using to track American credit card data (it paid Argus $2.9 million in 2012). That covers pretty much everything else.
A Log of What You're Watching on TV Is for Sale, Too
But of course, none of this data is valuable to advertisers without viewer statistics, and that's where the cable companies come in. Some have been twitchy—one cable exec begged me two years ago not to print that they were doing anything with their set-top data (they weren't then, but they are now) because the company was worried about getting called in front of Congress. It's a very specific data set, and with most cable providers, it's opt-out. On the back of your bill, usually, there's a little box you can check and send back in to your cable provider saying "please don't use my data for marketing purposes," but if you didn't know it was there, you probably didn't see it and you're probably participating completely by accident.
The data is everything you do with your cable box—what shows you watch, whether you watch them time-shifted or live, and which ads you see and which ones you skip. Granted, some companies aren't that sophisticated in terms of what their own personal box does, but the most impressive of them—Comcast, for example—definitely do.
This data is cross-referenced with all that information on how you spend your money that Acxiom, Experian, Argus and others provide. It's anonymized, but it's anonymized in a very sophisticated way—the company that provides ROI data buys both the viewer information and the financial information and then compares the two down to the individual address to make a single "pod." Using your address, it matches your Kmart and CVS loyalty cards, your car registration, and your credit card data with your cable bill—they all go to the same place, naturally—without looking at any of it, and then strips out your name, your house or apartment number, and your street. Then it looks at the data.
Several Companies Use This Data To Make Marketing Money More Effective
AOL recently bought both PrecisionDemand—which was folded into Adap.tv—and Convertro, both companies that deal with this data. Why are they so valuable? Well, because a simple Excel spreadsheet won't cut it—not with data sets this complicated.
Now all these companies know is that there is a person living in your neighborhood who buys and watches the things you buy and watch. They don't know it's you, exactly, but they almost know it's you.
But Networks and Brands Actually Do Use Your Personal Information to Make Ads Less Awful.
That information is incredibly valuable to marketers, and they can do amazing things with it. By looking at it, a company like Adidas can say, "Hey, these nine blocks in the Bronx really love Nike. Through one of Nielsen's tools, let's take out a targeted buy through their cable provider and ask Facebook to sell us traffic stats on sites those guys like." When they've done that, they can look at how the buy did, and see that few people who saw the ad during the day cared much about it, but almost everybody who was watching Adult Swim at night and saw their ad went out and bought a pair of their new shoes over the next three weeks. That makes advertising many, many times more effective.
Jeff Zwelling, CEO and co-founder of Convertro (which was a part of Nielsen's Multi-Touch Attribution pilot program—we reported on that last week), explained that Convertro doesn't use all of the data sets mentioned above, and some it gets at other ways. But the basics of the Convertro program are worth understanding, because they're very solid and likely to inspire other ad companies trying to do this same thing.
"Let's say in a perfect world where there's only one medium—let's make it a display campaign—we have a thousand people exposed to the display campaign and a thousand people not exposed to the display campaign, and half of the people who saw the ad behaved differently from the half who didn't," he says. "Getting people who weren't buying to suddenly buy is the goal."
And whether or not that goal has been achieved can be measured extremely effectively: "We'll say it costs $1,000 to expose 1,000 people, and when half of them change their behavior, we'll say the changeover is worth $5 per conversion," Zwelling said. "So for 500 people to change their behavior, that generates $2,500 of revenue. It cost us $1,000 to get there, so the ROI, or ROAS (return on ad spend) is 2.5—$2.50 for every $1 spent."
That's the dream model for advertisers—direct knowledge of exactly what effect your ad had on the group of people who watched it. "The problem," Zwelling points out, "is that in the real world, isolating consumers so that some are exposed and some aren't exposed, and isolating colinearity—whether someone was exposed to a billboard or a magazine—is impossible."
Zwelling and Convertro get around that by boiling down historical data to track ad consumption and purchasing; then, when, say, International Clown Nose buys their data, Convertro predicts how many noses will get sold based on that data, places ICN's ads accordingly, and tweaks its predictive data based on nose sale information that comes back in a few weeks.
Of course, the ads for the new, deluxe clown nose were really trenchant (people cried in testing!), and that's not always the case. "When the creative is bad, [content owners] don't want to be blamed," Zwelling says. "The biggest campaigns in digital advertising we've actually seen publishers spend money to improve creative. And we calculate what happens when you put an ad in front of Midwestern moms instead of coastal hipsters."
This also has the effect of separating the wheat from the chaff: this kind of data is expensive, and the companies that can't afford to use it will get left behind. You can see this happening on the Internet right now: if you opt out of Google's tracking program, the ads you get delivered are the absolute dregs of the creative world. Truly disgusting ads for skin creams, wart removal, scary-looking supplements, questionable weight loss solutions, and so on. The better experience for someone who browses the Web is unquestionably the tracked experience—you see ads for books you stand a chance of enjoying, movies you might like, and clothes that would look nice on you. Within a few years, the same will likely be true of television as directly targeted, or "addressable" advertising gets more sophisticated.
The problem is, if you don't like the idea of companies that want your money constantly looking to see what you're buying and tuning their campaigns to get as much of your cash as possible, you may be totally out of luck.