One of my basic principles says „do not talk about politics or religion” but today I must “bend” it a little. Donald Trump won and right now he is The President of The United States of America. It is natural that he has supporters and opponents but I will not start this topic, do not preoccupy yourselves. Today I want to tell you how Donald Trump used Big Data, gave relatively little money for a campaign and won presidential elections.
First of all, we should make it clear what Big Data is. Big Data is a database of all electronic traces that we leave behind, i.e. likes on Facebook or Instagram, our Tweets, search queries, credit card payments, locations of our phone etc. These data have been collected for a long time and they still are. Some time ago they were used for not too outlandish targeting advertising, but right now this situation is changing.
It is a field of science seeking to measure human personality. Basically, we can say that the goal of psychometry is to be able to say exactly what type of person we are dealing with, i.e. which needs and concerns they have, and what is their normal behavior. What does it have to do with Big Data? Michal Kosinski – a polish student who was admitted to Cambridge noticed that personal data collection performed by old methods is very laborious because researchers have to send thousands of personal questionnaires to people, with many questions each. He proposed to use Big Data. Quoting an article from Das Magazin “Kosinski proved that with just 68 Facebook likes, it was possible to predict skin color (95% accuracy), sexual orientation (88%), and their political affiliation (85%)… but it didn’t stop there.”
Model created by Kosinski is constantly being improved to soon based on 70 likes get to know the person’s character better than a colleague would, after 150 likes better than parents and after 300 better than life partner. As it is not difficult to guess, this method was quickly noticed, and thus Kosinski received an offer of cooperation. He rejected the offer because the company proposing the cooperation was SCL – a management agency dealing with elections around the world.
It is a company connected with SCL. You could hear about their activities in 2015 when in the UK there was a referendum on leaving the European Union. Fraction supporting Brexit revealed later that Cambridge Analytica has supported their online marketing. The method they used was similar to the Kosinski’s model, whereby the researcher needed to explain that there was no cooperation between them. Ten months later Alexander Nix – CEO Cambridge Analytica – joined election staff of Donald Trump.
The previous election campaigns looked very similar. The same message hit all people or the messages were personalized by gender, race, orientation, place of residence etc., generally speaking, adapted to demographics. In marketing, everybody knows that the personality of the client is the key. Putting personalized advertising is more effective and cheaper than advertising demographically. Alexander Nix proposed this way. Nix’s company bought and collected personal details and information about US citizens. In this way, they had a database of more than 200mln voters.
The entire database was divided into groups in terms of personality, which allowed them to adapt the message about a specific problem to a particular group of people. Let’s take an example of possession of weapons. For a neurotic and conscientious person, the most efficient will be a rational message based on emotions, e.g. fear of robbery, where a gun may be a form of protection. For closed and conciliatory people the best will be to base the message on sentiments, e.g. mentioning father teaching a son how to hunt, the same way his father and grandfather had done before. A properly prepared message is later sent to appropriate people by email or advertised in social media.
With this method, Trump’s defects become in a sense his advantages, and it’s all because his staff could create a separate message for almost every voter. After the 3rd debate, Trump’s staff sent to voters more than 175 thousands variations of his arguments. You can ask whether this kind of marketing is a fraud. In my humble opinion, it is not because Trump used a technology that was available, but of which his opponents did not know. Clinton gathered more votes, but votes for Trump were precisely distributed.
Nix said that they received for their services 15mln dollars. This is actually a drop in the bucket when it comes to the standards prevailing in U.S. Trump spend over half a billion dollars less on the campaign than his opponent because they specified his message to voters.
Cambridge Analytica does not disclose the impact they had on the outcome of the election, however, other countries have already contacted them with their offers.