Content vs. cookie: Which one tells us more about you?

Is there life without third-party cookies? (Part 1 of 3)
Conventional wisdom says the cookie is much more valuable than content analysis for telling us who the reader of a web page is. If you dig into the data, you can see that this is just plain wrong.

You probably have no idea how many third party cookies are laid on top of every piece of content you consume in your Web browser and how much they determine what you see — especially which ads you see. These cookies are from other companies besides the one whose website you are currently viewing, but they have the blessing of that website, and can be used for tracking purposes. Last year, a study showed that 86% of major publishers lay third-party cookies on top of their content for you. For example, according to CookieCert, which aims to be the largest independent database of cookie information, TMZ.com sets 150 third-party cookies (most browsers allow up to 300 cookies at a time).

Credit: Amy/PositivelySplendid.com
Credit: Amy/PositivelySplendid.com

BlueKai’s third-party cookie reportedly shows up on more than 48,000 different websites and you have at least a 24% chance of picking up a BlueKai cookie on a typical, random Web browsing session. BlueKai is a good data management platform (DMP) for us to look at, because they are exceptionally forthcoming to the public about what they are doing. In fact, they have a registry where you can look up, at any given moment, what BlueKai has recorded so far about you for targeting purposes.

Hats off to BlueKai for making this registry visible, and for allowing me (or you) to delete specific info or to opt out entirely.  And it’s rather impressive how many accurate characteristics BlueKai has recorded about me. They know I’m a technical professional who completed grad school, now living in a single-family home in the Bay Area.

On the other hand, the profile erroneously shows me to be a Spanish-speaking new mother. I can tell you that 99% of the time, I’m the only one using my laptop. It’s amazing to see how that remaining 1% allows these errors to creep in.  I don’t speak Spanish, but some in my household do.  My youngest child is 13, so forget about being a new parent. However, I shopped — just once in the last year — for a baby gift for a friend. Is that what fooled them?

There are also several pairs of items in my BlueKai registry that contradict each other: I am both male and female, both introvert and extravert, and my combined household income is somehow in four different brackets at the same time.

When I used a different browser, I effectively activated an entirely different tracking identity, whereby BlueKai had a completely different impression of me. For example, in this second profile, I’m merely a college graduate — no evidence of grad school. And it has different errors, too, such as showing that I have grandchildren (hey, maybe someday!).

I actually think BlueKai is doing about as good a job as anybody can — based purely on cookies. They can’t do anything about the fact that a cookie merely tracks one among several of my browsers and that a browser can be shared by others.

Let’s compare this with what you can infer from the content that I am reading (as well as where and when I am reading it) without looking at any cookies. The first thing I read yesterday morning was Kevin Acee’s analysis of the Sunday Night Football game. His article was written at a grade level of 8.6 and mentioned the San Diego Chargers and the Oakland Raiders. In reading it, I was accessing a San Diego newspaper website from an IP address in Northern California.

From all of the aforementioned information, we can infer that I am a sports fan, a football fan, and I may have an interest in domestic travel (given that I was reading the sports section of a local newspaper in a city that is far from where I was currently located). That’s already enough information for me to be shown several relevant marketing messages. Even though the information doesn’t determine whether I live in San Diego and am now visiting the Bay Area, or the other way around, an ad for Southwest Airlines mentioning San Diego would be equally relevant, either way.

Next up on my morning reading was an article in The Economist, at a grade level of 9.1, discussing the impact that the US government shutdown could have on markets. That may not seem like much of an increase in grade level over the last article, but combined with the fact that the Economist generally comes in at an average reading level of 10.3, it implies that I am in the top 25% of Internet readers, and likely college educated. The article I selected is full of pessimistic language relating to business and the markets. This is telling you a lot about where my head was at for the space of ten minutes yesterday morning.

The combination of a likely higher education level and an interest in finance, implies all kinds of things about what marketing might interest me — perhaps ads for financial products and other publications that are related.  I also showed interested in problems and warnings, so a marketer could even infer how messages in the ads should be tailored, e.g. ads mentioning financial uncertainty in the future might fit best into the frame of mind I adopted while reading this article.

All in all, it seems that looking at the content I am reading with respect to time, place, theme and mood, is at least as promising as what was dragged along by my cookie. Meanwhile, immediate contextual analysis does not creep me out in the way that building and storing a permanent profile of me does. The context analysis is right there in the current Web page and can be had in the absence of a third-party cookie. Better still, it’s relevant to what I am reading or browsing right now, as opposed to what my daughter did six days ago when she borrowed my laptop for five minutes.

Cookie tracking and content analysis are not mutually exclusive. There’s no reason we can’t use content analysis together with cookies to determine what would best match a user’s interest. But thinking that the cookie is a be-all-to-end-all is simply false. And when third-party cookies are not present, we’ll have to rely even more on the content analysis.  In the next post of this series, I’ll explain why third party cookies are destined for a dramatic decline.

Leave a Reply

Your email address will not be published. Required fields are marked *