What do they really know
This article discussed the targeted advertisements made available to modern industries, and the information made privy to the industries. By tracking our movements online, from the websites we visit to the purchases we make, different companies like retailers and credit card companies collect data on us, and sell them to larger data companies, creating enormous profiles on consumers to narrow down their preferences to a list of attributes – numbering up to 30,000. Then, based on this, advertising services sell advertisements to different willing industries to appear to consumers who match their desired keywords. As with the advent of “pay-per-click” advertising as discussed in lecture, advertising services would need a good reassurance that they aren’t just filling adspace without gaining revenue, so it is in their best interest to make the advertisements as targeted as possible to ensure maximum interest from every consumer to every ad.
The New York Times sought to expose some of these inner-workings of keywords and attributes by putting out targeted ads with the text informing the readers of what keywords were used to put the ad in front of them, displaying just how specific of a group an ad could target. These keywords spanned a wide range of descriptors and attributes. Some were more personal-life and habit-related, and while invasive, their uses for retail purposes were clear – such as “trying to lose weight”, “high spender at luxury department stores”, and “love bakeries”. On the other hand, others hinted at different uses – “registered Democrat” and “likely to vote for the sitting president” had a clear political leaning, and indicated at uses for targeting information towards people of different political leanings, much like the accusations during the latest election. While we can’t be sure what the prices of these ads were, we can be sure that they followed the laws described in class.
There is a caveat though – the data wasn’t always as definite as it may have seemed. Hinted by the “This ad thinks […]” format of the New York Times’s ads, much of the keyword-consumer matching done by the advertisement companies was done through predictive technology rather than hard data. One company described their process as “birds of a feather”, where they associated people with attributes based on attributes that tended to go together. These predictions could backfire though, as shown by the 2018 study exposing data brokers for attributing gender with readers less accurately than a coin flip.