Published by Dutch Journalism Fund (Dutch)
We have been able to customize sneakers and bikes for years, but the personalisation of news is still in its early stages. By analysing user data, it became possible to adjust online coverage on an individual basis. What initiatives are already being implemented, and which risks arise with personalisation?
More and more companies offer the option to personalise their products, and news organisations are also encountering a desire among readers for personalised coverage. This is no surprise when you know that an average of 36% of consumers have an interest in personalised products or services, according to research from Deloitte from 2015. The one size fits all model seems to connect less and less with the current generation of young consumers. Or like the insurance slogan says: “Why pay for something you do not use?”
Personalisation of news does raise questions. Is it, for example, necessary to keep online readers happy? Which initiatives for news personalisation are already in place? And is it ethical to spread news unevenly amongst people? To answer those questions, we should first define what personalisation is.
Friends amongst each other
A website that shows each user a unique homepage based on data about individual preferences, uses personalisation. According to the Digital New Report 2016, in recent years Reuters Institute news organisations have been experimenting with offering ‘specialised news based on preferred subjects of readers themselves, or in some cases, automatic generated recommendations based on user history’.
“Personalisation of content was almost impossible in the offline era. Now it is much cheaper and easier.”
Offline personalisation was almost impossible; creating, printing and distributing a unique newspaper per customer was not affordable. With the transition to digital news personalisation, it is a lot cheaper and easier now. Before you had to conduct questionnaires and interviews to map out the preferences of every reader. Now you can simply monitor who visits your website or app, and what they consume. The used device, internet browser, time and search history are examples of data used for personalisation.
Storm in a glass of water?
The question remains about the importance of personalisation in media land in order to keep users on the online portal. This becomes a lot clearer when we look at the way users find and consume online news. According to Natali Helberger, professor of Information Rights at the University of Amsterdam, 30% of adults in the United States consume news through social media. Besides that, 80% of traffic to news websites comes from social media and search engines, according to Stéphane Cambon, CEO of OwnPage Technology; a company that offers personalised newsletters to publishers. Now only 1 out of 5 visitors ends up on a website because they specifically type in the web address or click through from a newsletter.
This means that people are looking for information about a certain subject online; not for information from a certain news organisation. Distribution of news is therefore only partly in the hands of the news organisation that creates the news.
News organisations only have limited control in attracting readers. Personalisation of content, for example by using personal newsletters, is a way to attract readers and, once they are on the website, keep them there by showing targeted messages. Afterwards you can increase your potential income from paid content and advertisements.
Because of the increasing possibility and necessity of personalisation, different news channels have already started experimenting. On the Dutch market, Blendle is a well-known example as it adjusts the offers more and more based on what it knows from the reader; for example, the fact they just boarded a train. The BBC offers personalised news through the BBC News app. Besides categories like Top Stories and Most Read, the app also provides My News stories that connect to the interests of the user. This content can be compiled based on the preferences users give and similar stories are suggested based on related tags.
“An article about gay couples can lead to a whole load of wedding messages.”
The New York Times also uses tags and user history, but adds another method to decide which articles to show each user. According the New York Times, the problem with tags is they can give inaccurate results. An article about gay couples, for example, can lead to a whole load of wedding messages. That is why they also look at the articles similar users have read. This method has its own pitfalls (for example, it skips new, undiscovered articles), but the combination of both methods seems to work well.
Experimentation is not limited to online content; the first personalised print editions are also being developed. Australia Regional Media started with a custom-made advertisement appendix in the newspaper. This opens the road to a very personal newspaper, but like Editorial Director Bryce Johns says to New Media Works, ‘people should be prepared to pay for it.’
All these initiatives carry ethical questions surrounding personalisation, because what is the danger of making certain messages harder to see or more easily visible to an individual? According to the Reuters Institute this blind spot, or ‘algorithmic discrimination’, is a definite concern. For example, poorer people, who barely receive financial news, are mentioned. Personalisation can also lead to a stream of similar news called filter bubble.
Upday is a start-up that tries to prevent unwanted effects of personalisation. First, Upday shows you articles based on your preferences, but every now and then it adds a surprise to your list to see what else you like. Besides that, it also shows specially-selected choices from a team of journalists. According to the Upday website, the editorial office team makes sure important information cannot be missed.
Of course, another concern is whether readers want their, sometimes sensitive, interests shared with a third party. Natali Helberger points out the danger that political and religious preferences are shared with ‘American intelligence, Google Analytics and Adidas.’