I recently accepted an invitation from Acxiom, the big data broker, to check out some of what they knew about me on a new web site. What I discovered was a considerable surprise; in an era when both private data aggregators and the government are presumed to know everything about us, most of the information Acxiom had compiled about me was simply wrong. If my record is typical, marketers who are paying for this stuff are throwing their money away. They’d be better off with random guesses.
You create an account at AboutTheData by supplying some basic information about yourself–full name, address, the last four digits of your Social Security number. When you return to the site, you need supply only your email address and four-digit SSN. This would be a horrible privacy issue if the information that is exposed weren’t completely useless. AboutTheData breaks information down into several broad categories: Basic demographics, home ownership, economics, automotive, shopping, and interests.
Beyond democraphics. The demographics were fine, not surprisingly since this is easily obtained public information including age, education, occupation, and voter registration. Giving my race as “other” was bit odd, but I can live with that. The home ownership data was mostly OK as well. It basically is the same stuff you can find on Zillow, with the same minor errors (which are mostly errors in tax records.)
But things go seriously off the track in the economic data. Our household income was off by an order of magnitude and was seriously inconsistent with other data, suggesting a complete lack of sanity checking of the data. It listed my credit cards, including the American Express card I have not had in at least 25 years. Then it listed my number purchases five for cash, 15 on credit cards, and five “other.” It did not specify a time period, but that’s too much for most days and not enough for most weeks.
The automotive section was just weird. It had a correct date for when my next car insurance payment is due, but no information of vehicles. The shopping data was more garbage. The amounts spent and number of purchases–12 online and 11 offline–were again ridiculous. The “Most Frequent Retail Purchase Category” was “Standard Specialty; Specialty Apparel.” I’m not sure what that is, but it’s wrong. We don’t buy a lot of clothes. And in a checklist of purchase categories, every type was marked “true,” but the categories are so general that most of them are true for most people.
Interested in everything. Acxiom also marked “true” for all of a very long list of interests, many of which are not including cat ownership (dogs weren’t listed), hunting and shooting, boat ownership, and antiques. It said I was a modem owner, which I guess is technically true, and that I am a Windows user, also technically but incompletely true. I suppose the information Acxiom supplies to real customers is better, or at least more detailed in its inaccuracies.
The company does stay in business and its customers can’t all be total idiots. But KnowThe Data raises two possibilities, both really disturbing. Perhaps the data they have chosen to reveal is deliberately obfuscatory, in which cases the Federal Trade Commission and states attorneys general ought to be coming down on them real hard. Or the information they sell really is terrible. The latter might at least explain why I keep getting marketing pitches for products that I would never consider buying.