Facebook earns phenomenal income out of advertisement targeting tools based on the data and metadata generated by its users. We all know that. As an anthropologist, however, I am interested in how exactly this method of data-based targeting works. Because behind it, there must be a kind of worldview, a theory, if you wish, of how “normal” human beings run their daily lives, who they are, and what kinds of communities they form. A sort of Facebook “ethnomethodology”, in short, of considerable power and influence in today’s online-offline world.
There is, of course, no direct way in which we can examine this, for the algorithms defining the targeting strategies are among the world’s most carefully protected industrial secrets. But there is an indirect, ethnographic way.
On 2 December 2016, between 12 noon and 3PM, I did three long runs through my Facebook feed, and I collected all the advertisements it contained. What follows is an analysis of those advertisements. The assumption behind it is that the general patterns in these advertisements will reveal something of the underlying vision which Facebook, through its advertisement formulae sold to companies, has op people such as I – how Facebook constructs me as a person. This may not shed direct light on what Facebook knows about me, but it sheds light on one aspect of how that knowledge is enacted in this segment of its day-to-day operations.
In good ethnographic tradition, the materials need to be adequately contextualized.
-I have a main Facebook profile, which I maintain in Dutch. I use it as a news medium and a tool for what I call “knowledge activism”, explicitly on the left wing of the politican specter. This profile is five years old now, and has about 5000 “Friends” and about 1700 “followers”.
-I also maintain a “professional” profile page in English, mainly for international friends and colleagues, which as a half-joke I listed as a “nonprofit organization” with Facebook. It has about 800 followers.
-Facebook of course knows a few things about me. I presume it knows a lot about me, but have no idea what exactly is known. I am only certain of what I passed on, explicitly, to Facebook, in the way of personal information. Facebook knows, from profile information,
- my name
- the fact that I’m a man
- that I was born in 1961, which makes me 55 years old presently
- my location: Antwerp and Mechelen, both in Belgium
- the fact that I’m an academic with advanced educational qualifications
- the language I communicate in
I did not reveal my “relationship” type to Facebook.
I collected and screen-shot 56 advertisements in the three runs through my feed. Almost all of them are in Dutch, and in a first, rough scanning of the materials, three main categories could be distinguished:
- “Indiscriminate targeting”: Sponsored advertisements sent to me for no apparently specific reason
- “Friends’ interest”: Advertisements sent to me because friends liked them;
- “Age and biodata ads”: Advertisements of which I strongly suspect that they target “people like me” on bio- and user data grounds.
Let’s have a look at these categories.
Just under 25% of the corpus, 13 items, were sent to me through sponsoring.
The sample was taken in early December, in the days and weeks leading to some of the biggest shopping periods in my country: Sinterklaas (6 December) and of course Christmas (25 December). Sinterklaas is an event in which, to cut a long story short, children receive toys and candy, supposedly from a mystery saint (Sinterklaas), but in actual fact from their parents. The latter may explain the “Leo” ad – candy. But unfortunately, my children are adults and they know that Sinterklaas does not exist.
This is not the only ad that seems to be quite clearly mistargeted. The “Beauty Men” ad is obviously aimed at men considerably younger than me, the clothing advertised by “Wish” equally does not – to my taste – suit a 55-year old man, and the ad for an MBA study program at “HEC Paris” to make me “grow as a leader” also seems to address people in another stage of life and career.
The other ads are commonsense ones. Yes, I could be interested in saving fuel (provided I would drive a car, quod hardly ever), in cheap Christmas stuff from “Lidl”, in a new version of Google Chrome to run on a new Huawei phone, in wellness, cheaper electricity, flowers and design products.
This is the largest category, accounting for 50% of the corpus (28 items). All of these ads were prefaced with names of Friends who endorsed or liked the page now sent to me.
This category is obviously a bit of a dog’s breakfast. But I learn a lot about my Facebook Friends and followers, of course. Quite a few of them are deeply interested in art: there are ads for an art cinema, a publishing house, art and entertainment festivals and events. There appears also to be quite some interest in organic food, environmental issues, sustainable economy and climate change – several items make an appeal towards that (even if, weirdly, there is also one advocating nuclear energy). Quite a few of my Friends seem to be keen travelers and prefer the more adventurous off-road trips over standard packages or five-star resorts. And some fascinating gadgets are offered too – a piece of “spy tech” to locate my car and an amazingly sophisticated cycling helmet.
Age and biodata ads
This is the least clear-cut category – at least at first sight. I received 15 items (just over 25% of the corpus) in which obvious aspects of my “real” person appear to be targeted: my gender, my age, my (absent) relationship status. Several of these ads are also “Friends’ interests” – names of Friends preface the message – but an additional filter seems to be applied, in the sense that such ads would, I assume, not be sent to people with a profoundly different age, gender and/or relationship profile than me.
Things become a wee bit more scary here. I am constructed here as someone
- with several health issues
- who should start thinking about dying
- but is looking for love affairs
No less than five advertisements have to do with age-related health issues. Prostate problems, cholesterol, depression or burnout, diabetes and smoking are the issues which someone like me – a 55-year old man – might suffer of and seek treatment for. This is nicely punctuated by the funeral insurance ad, which reminds me that funerals are very expensive events and that my children should not be facing such crippling costs – rather soon, possibly (you never know).
An ad for a Mechelen-based chips shop provides a counterpoint to this theme of “take care of your health”!
There is, however, a bit of a Last Tango in Paris twist to the plot. Since I did not give Facebook any information on my “relationship type” – it is frankly none of their business – the machines appear to conclude that this 55-year old Belgian, whose body is subject to decay, might still be interested in love affairs. Perhaps even more than ever, and perhaps even in an “it’s now or never” mode. Two dating sites target my attention, one of which specializes in Muslim potential partners.
The world, and I, according to Facebook
If I would be asked which of these ads would really interest me, I would be hard pressed to choose one. I could take a closer look at some of the art and culture ads from the “Friends’ interest” category, and I might be interested in some of the sustainable economy or climate change ads from that category too. But none of it I find compelling.
This is even less the case in the “Age and biodata” category. I do have adequate health support, I do not need to be reminded of the fact that I am mortal, and I do not need any support for my love life either. My children might be planning to buy me the “1961 Legends” T-shirt for Christmas, but not me. I stopped using the services of Amazon France and my “business” (the “nonprofit organization” of my professional profile page) has no need for additional IT support.
So, who is this person constructed by the advertisers’ algorithms on Facebook? Here are some elements of an answer:
- That person has been constructed on the basis of mainstream behavioral characteristics: he is someone who does shopping and looks for “normal” shopping items: smartphones, car-related items, electricity provision and so forth.
- His shopping habits are, on the one side, defined by statistical averages drawn from profiles. Since he is a 55-year old man in Western Europe, he might have cholesterol, diabetes or prostate issues, and since he has not rendered any information regarding his relationship status, he might be the kind of man who goes dating.
- His habits are, on the other hand, defined by an assumption about the Facebook community built around him: this community is seen as a community of interests. Or to be more precise: a community of consumer interests.
To develop the latter: the Facebook advertisement system appears to presume that its communities are made up of people who, by and large, share the same lifestyle, sociocultural and socio-economic features, and sociopolitical value orientations. All of those forms of sharedness, subsequently, can be converted into consumer preferences.
Since my use of Facebook is, clearly, deviant – I post hardly any personal information and use my pages for the public purposes I specified above – the match between my community and myself is highly unclear, nonlinear. My community is very large and heterogeneous, with people of different age groups, gender and ethnicity, geographically dispersed over Belgium and The Netherlands, not homogeneous in terms of educational qualifications and professional status. The exercise I undertook here, therefore, teaches me more about specific Friends than about myself, since my use of Facebook constitutes a sustained “breaching experiment”: I appear to violate several key expectations of the “normal” Facebook user.
At least, it teaches me that Facebook, in this aspect of its operations, constructs me through the lens of my community. I must mirror some of the characteristics of the people who are my Friends. Since my highly diverse community counts a good number of Muslim Friends, I must be potentially interested in a Muslim partner-in-love; since several of my Friends are unemployed and struggling to make ends meet, I must be interested in cheap and second-hand shops as well. Since many of my Friends have a car (and talk about it or “like” car-related pages) I must have a car too and share these interests. And since rising fuel and electricity bills are a major topic of concern among middle-class house owners in Belgium, many of whom are Friends, I must be interested in lower fuel and electricity rates, of course.
Facebook, as we can see, uses a pretty simple , schematic and linear worldview in its algorithmic strategies. It is based on commonsense and statistical understandings of “normalcy” – everyone is designed as an Average Joe within his/her data-based category. It also uses a simple, schematic and linear view of human behavior, since these forms of normalcy are precipitated into equally “normal” consumption habits and preferences. Humans are essentially consumers, and when they congregate on Facebook, consumption must be a key interest; it must even be the determining interest that brings them together as a community. There is a disconcerting underlying assumption here: that Facebook is made for like-minded, highly similar people congregating in like-minded, highly similar groups. A rather suffocating, even disabling view of human communities, in my view.
It’s all quite amusing, were it not for the fact that this is one of the many identities I have. I have not constructed it myself – Facebook’s advertisers’ algorithms give me this identity. And whether I like it or not, it is a real and effective identity: I will continue to receive ads for funeral insurance, cars and car-related gadgets, Muslim date sites, cholesterol pills and so forth. For Facebook is very unlike to change its mind about me. For Facebook, this is me.