In the data world, the topics of transparency, privacy and ethics are more and more discussed. As we predicted a year ago, this may well be the next revolution in data science. We ran an experiment to understand the level of information Facebook users get on targeted advertising.
Data literacy is essential
Educating people on the role of data in their lives is essential. We see too many companies which revamp their terms and conditions and think they are done. Most people don’t read T&C and firms, if they are serious about being fair with their customers, should do much more than just changing the little lines on their websites. To give just one example, the recent changes brought to Twitter terms and conditions were certainly a good step towards a better understanding of how personal data is used to personalize their services and make money, but it is still insufficiant.
Targeted advertising on Facebook
The king of data exploitation and monetization is certainly Facebook. It’s the paramount example of data collection, analysis and monetization. No private company on earth has accumulated such a vast amount of very granular and intimate data on so many human beings. These insights are used to enable targeted advertising. Most people don’t know about the origins of the ads they are seeing and Facebook doesn’t really make an effort to provide them with explanations.
All Facebook users can get information on how they were “chosen” for personalized advertising. Click on the drop-down box on the right hand side of each ad and you’ll have the opportunity to get some explanations. Unfortunately the latter are very concise, too concise actually to be informative and reveal which personal data is actually used.
Two very different profiles : two very similar targeted information
Consider this experiment we ran on 3 different computers, each one associated to one different, specially created, Facebook profile. All 3 profiles had the same exact list of friends. One profile didn’t interact with anybody (it didn’t like, comment or share anything). The other one (the “high-end” profile) liked pages of high-end holiday destinations, commented on posts related to those destinations. The last one (the “low-end” profile) liked low-cost destination related types of facebook pages and content.
Next to the very different ads and contents recommended by Facebook’s algorithm (that’s a topic for a dedicated article), we see that the information provided on very different ads for very different profiles doesn’t vary much. Basically, all you’ll get is some basic information about the age categories and locations targeted. This is obviously neither sufficient to get an idea about the actual criteria used by the advertiser to target you, nor to understand what Facebook really knows about you.
Conclusion : is Facebook transparent about targeted advertising or not ?
The information provided by Facebook on its targeted advertising practices is much too evasive, inaccurate and doesn’t enable users to understand how their personal data is collected and monetized.
This is really pity because Facebook does have the means to educate people and make this world a better place, more respectful of privacy and individual rights.
The 21st century must fix the issue of ethics in data science or there won’t be a 22nd one.
Posted in big data, Marketing, Research.
