The Key Ingredients for Awesome UX

Now that we know a bit of history about UX (Part 1), and that we understand how our brain processes information (Part2), let’s move on to the process of UX, and understand why data is important in UX design and decisions.

UX and Data

We talked about different levels of processing information and why understanding these levels is important for UX designers. Another big chunk of the UX process is understanding data.

A vast majority of UX decisions made are based on data, and good design is always driven by data. Data gives us insights into behavior, helps uncover issues, and fixes existing problems. Data can help us confidently predict user behavior and explore new opportunities, but it can be very challenging as well.

Every company collects vast amounts of data about their customers, their behavior, and their patterns every single day. The biggest issue with this data is that, oftentimes, it becomes just interesting numbers that lack any actionable insights. How do we make sense of it? Asking the right questions is the key—what are the problems we want to solve, and which metrics do we need to benchmark and track in order to address these problems?

Types of Data

Generally speaking, we have two types of data: Quantitative and Qualitative.

Quantitative data is anything that can be measured by numbers. Much of today’s data flows from analytics platforms—how many website visitors did you have, how did they get there, how many people clicked on a given button, what is the conversion percentage, how many abandoned their shopping carts, and so on.

Even the most organized sets of quantitative data don’t answer all the questions about UX: how did the product make them feel? Why did they take a specific action—or not? What were their expectations and were they achieved? This is when we need Qualitative data.

Qualitative data is collected through interviews, surveys, usability tests, etc., which can also be measured. System Usability Scale (SUS) and Single Ease Question (SEQ) are some of common usability testing techniques.

Generally, what people say they do and what they actually do is very different. When we compare online studies vs. in-person interviews and tests we see substantial differences in how people perceive products and services.

It’s crucial to look at both qualitative and quantitative data to make well-informed design decisions.

UX Process

UX process can vary from company to company or from product to product, but generally, the very first step is identifying issues. Once the issues are known, the next step is to articulate the purpose—what are we trying to achieve, and is it an issue worth solving? What measure do we need to increase or decrease?

We need to critically evaluate those issues and if they have a real purpose. There are plenty of products today that “solve” problems which don’t really need solving.

UX3_1

Third, successful UX processes examine analytics and big data. What is the user’s behavior, where do they come from, how many convert and how many don’t, how many click A and how many click B, and so on. Once there is an established understanding of some data averages, we can start looking at potential causes of the problem we identified. That’s when interviews and usability tests are conducted to discover how people are actually using the product, how it makes them feel…frustrated and annoyed if your product sucks, or happy and amazed if your product is great!

Having looked at big data and identified potential causes, we move on to first iterations—low fidelity mock-ups and wireframes, followed by early testing to validate our hypotheses. We do as many iterations as we need, followed by more testing, depending on the complexity and time we have. Once we are happy with the result, we move onto visual design (probably a few more iterations there) and finally development.

This is, of course, a quick and simplified overview of the UX process.

UX Process and Norman Door

Let’s illustrate the process we just outlined with the Norman Door. The Norman door is a probably the most famous example of a product with design elements that give off the wrong usability signals….have you ever pulled a door when you really needed to push?

Issues:

  • People get frustrated and annoyed
  • Wastes time and energy

Purpose:

  • Design a better door
  • Increase efficiency
  • Decrease errors

Analytics and Big Data*

  • 70% of people get it wrong all the time
  • 90% of people are not satisfied

(*These numbers are totally made up. But they probably aren’t too far off.)

Potential Causes (Qualitative)

  • Conduct interviews, measure SEQ score, find out why people aren’t satisfied
  • Memorability – impossible to remember
  • Learnability – impossible to learn
UX3_2

First Iterations and Testing

First sketches, low fidelity wireframes and testing is then performed.

UX3_3

More Iterations, More Testing

Then we do as many rounds of iteration and testing as we need, considering complexity of the application, budget and timeline. When we are happy with the result, when UX is efficient and error free, we move onto visual design.

Visual Design

Finally, once we are happy on how it looks, we move onto development.

UX3_4

Use Case: Netflix and House of Cards

Netflix, like most of companies, collects huge amounts of user information: what device are they watching a show on, what times, when do they click pause or rewind, when do they abandon a show, what genres are preferred, and more.

House of Cards was one of the first ‘experiments’ of that scale created by data driven design and was used as a test case for big data.

The show quickly became very popular on Netflix, in part because it was carefully crafted around the watcher’s tastes and desires.

For example, data showed that people who enjoyed the original 1990s show, also liked movies starring Kevin Spacey and films directed by David Fincher. (Guess who is directing the film and who is the main actor?)

Netflix have analyzed viewing preferences such as plot lines, to make sure the show doesn’t get too slow or too fast for their intended audience. They have predicted what users would like to watch and let their big data influence the final User Experience.

Was it successful?

Netflix committed to $100M for 2 seasons. Their subscription fees are $8.99 in the US per user per month, so in order to only break even, they had to get 500k new members. ­Since the show began, 17m new users subscribed. Of course, the causes of so many new members are not entirely influenced by House of Cards, but the series is considered to be very successful. (Read the full article on the General Assembly blog here.)

Data-driven UX

Peoples’ expectations are sky high and without good User Experience, we simply cannot craft a desirable product. Before designing a product, UX designers have a lot of homework to do.

Big data, or any kind of data without insight is useless. In order to turn data into insight we need to clearly define the problems we want to solve and determine whether they are problems worth solving. Successful UX takes a scientific approach to define those problems and drive design decisions based on all sorts of data.