Optimizing the user journey by looking into extreme users

Using relevant data about what users are doing on the product to identify improvements with the most significant impact.

CUSTOMER JOURNEY - UX - DESIGN THINKING - DATA ANALYSIS - CUSTOMER ACQUISITION - LIFETIME VALUE OPTIMIZATION - CONVERSION RATE OPTIMIZATION

The subject of this case study is a leading company in the digital signage content management software where I work as a full-time product designer. Solving problems and discovering opportunities that are not obvious and making them evident through a results-oriented, multi-method approach to design, based on empathy, data, and the ability to ideate quickly.

Problem

A digital signage content management system is the centerpiece and most important part of any digital signage software. It serves for the customization, scheduling, monitoring, and deployment of content to remotely managed displays.

We knew that users who succeeded at setting up their remote display in the first three days following registration were exponentially more likely to become customers. Few things played a more significant role in product validation as seeing the actual content on a remote display.

Although we converted a good fraction of our visitors to registered users, we kept failing at increasing display activations. At most, the conversion rate was standing still at 20%, if not going down. While the company was always looking for low-risk performance improvements, like A/B testing. I thought this was a particularly good time to move ahead as we have not seen many improvements in the last few years.

We needed a new user journey to get the big picture, one defined by a combination of channels, devices, and user tasks. Once we had this, we could develop a plan to prioritize changes with the most significant impact, while keeping risk to a minimum.

Tools and Process for Ideation

Finding the extremes and fundamental bottlenecks

The Design Thinking concept of looking into the extremes of a users experience helps to uncover fundamental bottlenecks. Fundamental bottlenecks are harder to fix, and harder to see because they are caused by some unknown fact of users nature or strong assumptions around the product (Bueno 2015). Without identifying these bottlenecks, it is common to fall into a pattern of fixing something that activates another bottleneck.

In this case, for example, increasing the number of signups and releasing improvements to the product did not move the needle, confirming the assumption that a fundamental bottleneck exists. Getting everything else right was unlikely to move the needle.

If you want to understand something, take it to the extremes or examine its opposites.
— Col. John Boyd

In my design process, I like to use models and algorithms for data analysis. I start by collecting relevant data, creating a dataset and changing constraints and see what shifts in response. Data alone won't do, I think the most critical thing in this process is asking questions while manipulating the data: What happens if I had to change the initial step for new users? Halved the number of steps? Creativity in the questions is also a way to design.

With this in mind, I set out to collect our product usage data from different sources:

Big Query

  • Which users had activated and when had they activated?

Web Analytics

  • Where were our users coming from? What were the interactions we had with them before the activation?

Zendesk

  • What questions were users asking at different stages of the journey? How were they affected by interactions with customer support? What areas of the journey caused the most people to drop and not activate?

This resulted into the following analysis:

Presenting the results

A proper design process is centered around efficiently communicating. Data analysis by itself is not enough for people to understand, believe and know how to act on the recommendations.

This is how I present my findings:


Implementation and Results

As part of this project I also developed...

 

Bueno, Carlos. “Shaping Big Data Through Constraints Analysis.” InfoQ, www.infoq.com/articles/shaping-big-data-through-constraints-analysis.