Let us cut to the chase. Facebook, Google and probably Amazon store everything you do in the digital domain. But any marketer who believes that true ad relevance is ever achievable should take a hard look at how the sites and apps they use classify them as individuals.
Here is what a recent report from the Pew Research Center has to say about the process by which companies figure out what site users might be interested:
“It is clear the process of algorithmically assessing users and their interests involves a lot of informed guesswork about the meaning of a user’s activities and how those activities add up to elements of a user’s identity.”
The report details the results of a survey focused on Facebook’s categorization of individual’s interests as represented by their “Your Ad Preferences” page. Not surprisingly, 74 percent of Facebook users did not even know the page existed but when directed to it 51 percent stated they were not comfortable with how the site categorises them, and 27 percent thought the categorization inaccurate.
Reading the report reminded me to check out my own Ad Preferences page. Assuming that the classification shown is an accurate representation of what Facebook thinks my interests are it is laughable. Apparently, I once clicked on an ad for Entre Rios Province in Argentina? Someone uploaded a contact list that brought me to the attention of a Land Rover dealer in San Antonio, a Toyota dealer in Colorado Springs and a Maserati dealer in Scottsdale? Oh, and I once liked a page related to “Life”. Yes, and I intend to enjoy it as long as possible.
Most of the categories are less than ‘informed guesswork’ and more literal interpretation. Take for example, the fact that I liked a page about “Yacht Racing”. Yes, I probably did. Why? Because my godson, Hugh, is a solo, deep sea, sailboat racer and I follow his progress with interest. It would not take a genius, or artificial intelligence for that matter, to realise that the only pages I have liked or viewed relate to events that feature Hugh, not yacht racing in general.
This problem of incorrect inference applies to all the data collected, including the behavioural data gathered from your mobile. My colleague Jane Ostler gives this hypothetical example of how geolocation can go astray,
“According to my mobile’s location data I am a regular at the gym. I spend half an hour there twice a week, and therefore see many, many ads for protein powder targeted at gym regulars. But my gym could be right next door to a donut shop, and I could in fact be stuffing my face with donuts twice a week.”
I have no doubt that behind the scenes Facebook can create a far more detailed and accurate profile of my behaviour and interests. The data is all there to be used. If an advertiser really wants to serve me relevant ads they should be taking all this information into account. But how many really do? From what I see, most simply take the basic classification and run with it.
Written by Nigel Hollis,Executive Vice President and Chief Global Analyst at Kantar Millward Brown.