Personalization was pioneered by Amazon when it introduced its book recommendation system (essentially a precursor of predictive analytics) in the late ’90s. Based on book purchasing history, a user was shown related book titles. Jakob Nielsen of the Nielsen Norman Group championed the book recommendation system as a success–users did not have to do anything, the system did all the work, and the system learned which books were similar by tracking users’ buying habits site-wide.
“Good personalization requires the system to know a lot about the user,” Nielsen wrote in 1998.
In 2015, behavioral, demographic, geographic, and device information can be easily obtained from a user. Using such data, usually collected unobtrusively, and implementing it well in UX design can inevitably “deliver a better, more relevant user experience, without being over-familiar or explicit,” according to Matthew Fiore, a writer for UX Magazine.
Personalization as a Trend
In our current culture of instant gratification, people want what they want, how they want it, and quickly. A 2013 Janrain study found that three-fourths (74%) of users (a sample of 2,091 adults) abandon websites which show content (e.g., ads) unrelated to their interests. With this study in mind, it is not surprising that UXPin deemed “Personalized UX” as one of the top UX trends of 2015 and 2016; no one wants to risk losing users.
Furthermore, more people than ever are doing their shopping on mobile devices, according to a 2013 study by comScore. E-commerce (or “m-commerce”) personalization (for example, allowing users to be notified by a push notification when an item previously out of stock is available to purchase) will only allow for significant boosts in revenue. “M-commerce” mobile app designers would be remiss not to use a personalized system.
All of the giants–e.g., Facebook, Netflix, YouTube, Spotify–implement personalization in some way or another. But interface designers are only just starting to discover innovative ways to play with personalization. Spotify, for example, is lauded by UXPin as having one of the “best examples of personalization you can find.” Runners can use the Spotify app to detect the “rhythm” or “tempo” of their running; Spotify then matches a playlist of songs with the runner’s tempo.
Perfecting the Art of Personalization: Evaluation
To test the effectiveness of personalized interfaces, traditional usability evaluation methods fall short, a 2007 study claimed. Traditional methods “rely on the assumption that presentation and content are the same for every user in every context.” There are new factors to contend with when evaluating personalization: for personalization to come into effect, a certain amount of prolonged interaction with an interface is required, thus asking for user investment. Secondly, evaluations need to use a diverse group of participants in order to detect the highest amount of usability issues.
User interviews, user surveys, user documentation, and customized journey maps, UXPin suggested in its 2015/2016 UX trend report, are ways to improve personalization. User documentation such as personas and scenarios are particularly beneficial. Personas represent data collected during user testing, and scenarios answer the question: how will users interact with your product to accomplish a task? and can streamline task completion. Lastly, a customer journey map, like scenarios, provides steps for task completion but with the addition of an outline of user’s “feelings and opinions” (watch a video on how to create a customer journey map here). In mobile UX, UX designers can use analytics (e.g., Appsee) and user recordings (e.g., real-time videos)–both methods track user behavior.
The Downsides of Personalization
Personalization is not infallible. It isn’t always accurate–we can purchase something from an e-commerce website as a birthday gift for a friend, and we’ll be given a list of related
items despite a lack of personal interest.
There is also the danger of too much personalization, as UXPin wrote in its 2015/16 UX trend report, which can repel users and make them feel uncomfortable. (“Much like in-person interactions, there is a limit to the intimacy people are comfortable with when first meeting”). This is why user testing is so important–you will be able to determine the level of personalization your users are comfortable with.
Having too much personalization can also lead to what Eli Pariser called the “filter bubble” in a March 2011 TED Talk. The filter bubble shelters us from expanding our worldviews and discovering new things; our provincialism could get the better of us.
Finally, the foundation of personalization is “passive data” (e.g., knowing a user’s location, past page views/purchasing habits, etc.), which means that information about a user is taken without his/her consent. This begs the question: are we willing to forgo our privacy, to any degree, in exchange for better experiences when using our mobile devices and desktops?