How predictive analytics will shape UX
- By Andrii Glushko
- Aug 15, 2016
When my 4-year-old daughter woke up the other morning in a bad mood, it took my wife only a few seconds to assess the situation, find the proper wording and voice intonation, develop a brief plan and start working to make our daughter’s morning better. She quickly recognized the need and understood what had to be done to fulfill it.
We’re often closest to the people (or even pets!) who understand us without the use of words. We like the feeling of not having to explain ourselves to be understood. Similarly, in order to properly marry user experience and predictive analytics in the future, UX designers must focus on interface personalization, context sensitivity and careful prioritization of information to be delivered.
In the scope of this topic, my daughter’s morning may seem unrelated; however, it perfectly illustrates the way we feel about someone predicting our needs. And once a product or service manages to evoke that feeling, we get used to it quickly and feel comfortable using it.
When we talk or read about predictive or prescriptive analytics in the digital world, we usually think about how data scientists are trying to make sense out of massive amounts of data. However, from the psychological perspective, the human need to be understood is driving the design of user interfaces with predictive features. We value longtime friends because we’ve put effort into getting to know them, and they did the same for us. We love them for hundreds of different reasons, one of which is that they are more predictable than strangers. We feel comfortable and safe with those who understand us.
The same principles apply to products and services that we use. People like using Google Now, Netflix and Spotify because they have put some effort into showing these systems, whether directly or indirectly, how best to deliver services. These services learn and become more predictable over time, and we tend to value their reliability.
A Pattern for User Experience Design in Predictive Analytics, written by Mike Kuniavsky, is one of the best pieces I’ve come across regarding the high-level rules that will help a designer stay focused on the right things while creating UI with predictive analytics. This short, simple and straightforward article suggests paying attention to three main areas:
- Expectations: Set expectations by making people aware of the predictive nature of the device, perhaps including information on how long it takes for it to learn or adjust.
- Explanations: Explain or display when a state has changed or a decision has been made based on the predictive algorithm.
- Affordances: Create clear possibilities for action in the flow of the experience to allow the predictive behavior to be adjusted.
What I would add here is making sure multichannel systems are robust enough not only to gather and process information from different devices and services but also to communicate with the users efficiently and consistently via different channels, depending on the user’s context.
As additional sources of data, physical sensors and software trackers supplement the system. As they become more responsive, they bring many benefits -- from an improved, overall user experience to the possibility of actually interacting with the system instead of merely consuming its content -- making processes more efficient.
The representation, order, amount and density of data must continuously adapt to a user’s current context of use, conveying just the right amount of information on the appropriate device.
Technological trends to watch
Apart from the overall trends that we as designers may see, let’s look at the technological trends that will drive our capabilities in the future.
There are many different tools available for data analysis as well as predictive and prescriptive analytics. Most of them are created by large corporations like IBM, SAP, SAS and Oracle, and they have fascinating capabilities. However, what interested me the most was the artificial intelligence company called DeepMind that was recently acquired by Google.
“DeepMind has been combining two promising areas of research -- a deep neural network and a reinforcement-learning algorithm -- in a really fundamental way,” the company’s cofounder and CEO Demis Hassabis said in an interview with Wired.
Although it is a different kind of tool, I find it more promising in potentially doing the analytics, decision support and assistance in the future, as the
The DeepMind algorithm can not only ‘learn’ its way through any new environment, but it’s doing so with hardly any entry data, simply having a task to start with. The results of this algorithm’s playing Space Invaders and Breakout games are very impressive and offer promising potential for providing analytics, decision support and assistance to UX designers.
Even though there’s a fear of trusting intelligent algorithms with serious decisions, I believe these tools will give us more human-like outputs, and designing for systems that will use those higher quality outputs will become much easier.
The same way a mother interacts with her child, algorithms will be able to learn about us faster and more efficiently, connect more dots and apply knowledge received from other domains to solve problems in entirely new ways. Our jobs as designers in this case would be to “humanize” those outputs, build correct expectations for the users and properly communicate the most crucial information needed for decision-making.
Although technical information is very important in terms of understanding a product or service’s possibilities and limitations, one of the most important qualities for the user experience designer will always be empathy.
Let’s go back and think about the 4-year-old. She’s sleepy and vulnerable and wants to be understood without explaining what’s going on. We all have that kid living inside of us, no matter what person, system or object we’re interacting with. It can be an online service that we use occasionally, or a complex and feature-rich application that serves as a main tool for highly skilled work. Both of them must analyze data to predict what a user needs must do, invisibly and gently, while allowing a user to easily correct the results of the system's work, if needed.
We’re about to see how predictive analytics and artificial intelligence drive UX/UI design to previously unimaginable levels of personalization. For the interfaces’ design, data visualization will start playing a key role as presenting more complex information in an intuitive and simple way becomes critical. With the Internet of Things and wearables increasing the number of devices providing services, companies must provide improved and carefully considered service touch points.
Andrii Glushko is a UX design consultant at SoftServe Inc.