Today’s customers have higher expectations of brands than ever before. A good understanding of the customer’s needs is necessary to reach these expectations. But how can you gain an understanding of their needs when you are dealing with a highly divers group of people through a medium where you can’t speak to them physically?

This post was originally published on the Edenspiekermann website on 10th of July 2014

At Edenspiekermann I’m currently researching the concept of personalized interfaces: a method to create personalized user-experiences. I believe that this method gives a great opportunity to tailor experiences — in a digital landscape— according to the different needs of customers. In part one of this series, I would like to explain more about the core of this method, the benefits and how it works.

The fundamentals

The concept of personalized interfaces is based on two fundamental ideas:

  1. The system is able to understand the user’s needs and adapt to them, making it possible for the user to reach their goal through a minimal amount of interactions
  2. The interface aims to predict the behaviour of the user

With personalized interfaces we aim to predict the user’s next step and use that information to lead them to an interface that suits their needs. But how do you predict user behaviour?

How do you predict user behaviour?

Predicting user behaviour is pretty much like looking into a crystal ball combined with statistics. It’s impossible to get all the predictions right. Take Google for example. Google Now tries to predict your next step but fails a number of times. Striking, if you ask me, as they have tons of information at their disposal.

The good news is that if we collect the right signals and use the technology at our disposal, we come up with the right prediction most of the time. With statistics and technology we have some effective tools to learn from human habits. Humans tend to do things out of habit: we do the same things day in, day out, and similar people do similar things. For example, if we go to work with public transport we always use the same apps to look for train schedules and delays. Or most people use Google to search the internet out of habit.

We can learn peoples behaviours through these habits and categorize them. Of course, I am simplifying a bit, because there are millions of people with millions variations, but you get the idea. The goal of personalized experiences is to define those repetitive actions, learn from them, predict them and hand them to the user in the most convenient way possible.

User satisfaction

If you predict users’ behaviour correctly, you can save them time and they will feel better understood. Also, this releases the user from the strain of processing boring and repetitive information and actions. You will see that when the users are satisfied with your online experiences they will become more loyal to your brand. And isn’t that what we’re all aiming to achieve?

How does it work?

In order to create personalized experiences we need to know ‘the context of use’. This can be determined by the core-functionality of a product, current state of the application, sensors in the device (GPS, Motion ect) and the next most plausible action of the user. These aren’t all the criteria that matter while using an interface; the criteria can also depend on the purpose of the product. But nevertheless, these are indications of ‘the context of use’.

Let me give an example of using ‘context of use’. In the Netherlands, bank debit cards don’t work by default outside of Europe. So, lets say for example, I’m traveling to the United States I can’t pay with my debit card until I unlock it with my banking app or through the web interface. When I log in through the web-interface from my bank when I’m in the states the bank could use the information of my location with the fact that I forgot to activate my debit card for outside of Europe. The bank could make a message on the first page I log in to which says: “ Activate your debit card for worldwide usage”, so I don’t have to search for it when I’m stressed because I can’t pay for my hotel.

In order to make interfaces personalized we need to build user profiles. These profiles can contain information, such as the following:

  1. What does the user click/tap within the app and how often? This gives insights into the user’s interests and preferences.
  2. How long does the user look at the content? If the user looks for quite a while at the content it implies that he or she likes the content and is engaged.
  3. What is the path the user takes? How does the user find the content they like? Can you make personalized “shortcuts” towards content?
  4. Let user make explicit decisions about what they like or don’t like. For example, Spotify radio lets you like or dislike a track. And refines the selection of future songs for the radio station.

Understanding ‘the context of use’ makes it possible to identify certain patterns in the user’s behaviour. But what if you don’t know this yet because you’re dealing with a first time user?

We can do this by asking the user for a little bit of input, relevant for your service. In the movie ‘Her’ directed by Spike Jonze you see an example of ‘first day questions’.

Examples of onboarding

In the movie the questions are unrelated to the goal of use, but the way of asking first day questions was good. With the answers you get from the first time user questions you can try to put the user into a group with similar people. And from there, you can build a profile. A good example of ‘first day use’ questions is Flipboard. When you use Flipboard for the first time you have to pick categories you are interested in. Flipboard uses this information to tailor the content to what you want.

In a concept air travel system we developed, we asked as a first time question the user’s booking number. With this information we already know which flight someone takes, if they travel alone or in a group and when they booked their hotel in the same booking also their final destination. If some information misses in the booking number we can ask the user the address of the hotel. But by only asking for the booking number we make it easy for the user to enter this crucial information.

In order to successfully create personalized interfaces you should always enable users to make choices within their own experience. Moreover, instead of forcing them to certain behaviour, it’s better to suggest actions and provide other options. We always need to give users an escape to do something else and learn form the escape they use.

In the air travel concept app, we also used the principle of ‘choice’. This feature helped passengers by giving them advice about how to get from the airport to their final destination. In this case, we knew already in advance that the passenger needed to take the train at a certain point in the journey. Instead of solely showing an option to go to the train schedule, we still gave the passenger the opportunity to open their flight ticket, luggage information and other features.

In short

By making interfaces learn about user behaviour you let the user feel understood, valued and cherished. And when you do this right most of the time, they will forgive you when you do it wrong.

To build profiles, and to put people in them, look for repetitive behaviour patterns and learn from those choices. If you have it right the user won’t notice. If you do this wrong too many times, or do it in an intrusive way, you will fail miserably (who remembers clippie?). So make sure you ask the users the right questions on the first day of use and give them choices in case your prediction wasn’t right so they can still find what they wanted to do.

This was an introduction to personalised interfaces. In the next post in this series we will go in depth into how to alter the way you display content to different user types, besides the functionality of a product.