This story is over 5 years old.

This Tool Shows How Your Email and Search History Gets Turned into Ads

Called "X-Ray," the tool opens up the black box of data that funds the internet.
It's all hell.

The internet is a big, black box when it comes to how our data is used by advertisers. Seeing a Gmail ad for something I just emailed a friend about doesn't much faze me anymore, yet I have pretty much no idea how my personal communications data got from the proverbial point A—as a keyword in an email—to its final form as a targeted ad.

X-Ray, an experimental tool that tracks how data turns into advertisements, aims to explain this relationship. Designed by researchers at Columbia University, X-Ray models the correlation between certain keywords, say, "debt," with the ads that return to the user, which might include something about car loans.


The vast bulk of the internet is funded by targeted ads, which use data mining to allow marketers to target us based on our recent searches, emails, and more. By trying to understand how our data is used, and not just collected, the Columbia researchers are hoping to bring more transparency to online ads.

To do this, the researchers sent out emails containing keywords both benign—"travel"—and sensitive—"AIDS." X-Ray then reported every ad that demonstrated a high degree of correlation to the topic in question in order to build a probabilistic framework that measures the relationship between the keywords and ads. Some of the results, which you can view on X-Ray's website, were surprising.

Emails and their correlated ads. Image: X-Ray

For example, the researchers found a high degree of correlation between the keyword "debt" and ads for credit cards and car loans. This finding is especially concerning, since it suggests that the ad firms representing credit card companies are targeting people already in debt for new loans.

Bafflingly, the keyword "depression" returned ads for shamanic healing and a texting coach site to help you score the girl of your dreams with one perfectly crafted message. While humorous, it's the ability to highlight exactly this kind of skewed, not quite 1:1, relationship between our data and the ads they engender that makes X-Ray such a notable tool.

While some services, like the Electronic Frontier Foundation's Privacy Badger, allow for some transparency regarding who is collecting data and how, X-Ray takes the next step to map how it's used. The tool is part of a growing trend among techies to try and give consumers some skin in the game in the online data marketplace.


the keyword 'depression' returned ads for shamanic healing

Telefonica, for example, is looking into a free market for data that will allow users to sell their information themselves to whomever they choose. Meanwhile, researchers at MIT have developed a data security framework called OpenPDS which allows users to keep their raw data in a black box of their own, before it gets into the hands of advertisers.

X-Ray is another link in the chain of consumer data empowerment, which seeks to explain what happens once users let their data go.

The point is that I don't know how the hell my conversations end up serving me an ads—only that they do. I'm left to ask, "Why me? What did I do to be targeted by an ad for a massage therapist in my area?"

While X-Ray doesn't have all the answers just yet—the researchers are keen to note that the service only maps correlation between keywords and ads, and not causation—it seems to be well on its way.