The news is currently filled with the large data analytics scandal of Cambridge Analytica and the use of Facebook data. The company developed predictive models which analysed the personal profile and usage behaviour and used it in an attempt to influence voter intentions. Here we see the ugly side of what analytics models can achieve, by classifying users according to their political orientation and targeting individual advertising campaigns. Software companies, such as Facebook, that host large amounts of very sensitive data have the responsibility to protect it. However, their business model strongly relies on targeted advertising and analysing it, and by third parties as well. A couple of years ago marketers could only dream of such granular targeting abilities, and nowadays an entire industry is built around online advertising campaigns.
There are two questions to reflect upon: First, what are the ethical issues behind using powerful analytical techniques to influence consumers and voters? Second, are analytical models already that powerful that indeed can provide such capabilities, or has this been exaggerated by companies and the press?
Surely many users appreciate when Amazon is able to recommend an interesting book, Spotify to provide the perfect Discover Weekly playlist on Monday morning and Google AdWords listing products instantaneously on search. There is a long debate by how much algorithms influence our choice already, but keep in mind that algorithms do not get it always right either. Sometimes it is even difficult to track the impact of marketing activities, and there is a long debate about what is the fair attribution of online marketing activities to sales. Sometimes, a good-old consumer review may provide much more useful information for a potential customer. Oh, hold-on who guarantees that these are not written by bots? Fraud (and fake news) detection is a fast-emerging topic and should be taken seriously. Welcome to the world of analytics! Like with many things, it falls to the user to apply such powerful tools responsibly. And this clearly has not been the case with the claims made for these political models – some target markets are easy to hit, but the truly valuable and unusual individuals are harder to identify than needles in haystacks.
“This week’s forecast” is a blog series, which covers commentaries, by members of the Centre for Marketing Analytics and Forecasting , about topical and controversial forecasts in the news.Back to News