“Who’s next?” Widget

Role

UX Designer
2023

Tools

Figma
Qualtrics
Ethnio

Skills

User Research
UX Design
Visual Design
Split Testing

The Opportunity

Ancestry’s homepage offered a rich variety of activity suggestion widgets. However, our analytics team discovered that most users viewed a small fraction of the total widgets available to them. Simply put, users weren’t scrolling much.

Moreover, several widgets remained pinned to the top of the page. This often led to users encountering a stale and limited set of recommendations. As a result, the product team brainstormed ways to encourage users to review a larger volume of activity recommendations on the homepage. 

Process

The team weighed several approaches ranging from introducing UI that encouraged users to scroll more to modifying the overall page layout to expose recommendations differently.

After some deliberation, the team settled on an idea I initially proposed: enhance the “Recently modified” widget at the top of the page.

This top performing widget successfully encouraged users to pick up where they left off. The team aimed to enhance this widget by introducing new types of recommendations that exposed fresh paths to family discoveries. 

The new feature aimed to increase the following metrics:

• Clicks per newer user
• Percentage of less experienced homepage users reviewing hint recommendations

We weighed several guardrail and informing metrics as well. The team hypothesized that introducing new discovery paths would increase these metrics while also leading users to more meaningful insights about their families.

Discovery

My initial questioning focused on which types of recommendations resonated most positively with users. I leveraged existing research to create a survey that asked users to rank several recommendation types.  

The survey revealed some key insights:

• When asked to rank the “Recently modified” widget in its current form, users expressed mixed feelings. Perhaps a fresh set of recommendations could generate a higher level of interest?
• Users valued direct line ancestors
• Users valued the ability to filter by their mother’s or father’s family line
• Experienced users expressed a desire to focus on “brick walled” areas of their tree
  

I leveraged these learnings in subsequent UI explorations.

Usability & Display

I led two rounds of lightweight user interviews sessions with a total of ten participants. I tended to speak with less experienced participants to ensure the new feature aligned with our business objectives. I structured the sessions to better understand if users:

• recognized and felt comfortable interacting with the new recommendation UI
  
• valued and correctly interpreted the new recommendation options

I devoted most of the interview observing how users interacted with the prototype while peppering them with targeted questions.

Several themes emerged from the sessions:

• Several labels confused users. Specifically, several users thought “Direct line ancestors” and “Near my tree’s trunk” referred to their immediate close family, such as their parents or siblings. In reality, the filter contained a much broader set of family members.
  
• Most users initially overlooked the new UI but most often correctly interpreted its purpose and function after noticing it.
  
• Nearly all users expressed a positive sentiment about the new set of recommendations and seemed enthusiastic about interacting with them.

After integrating this feedback, I prepared the validated designs for handoff to the development team. 

Solution

The team released an updated recommendation widget with several important improvements.

Firstly, the widget featured filtering options that connected the user to fresh paths to new discoveries. Namely:

• Last updated. Continue exploring the most recently modified people in the user’s tree. Functionality mirrored what had existed before the change.
• Recommended. Recommendations generated by Artificial Intelligence.
• Maternal line. Recommendations from the user’s maternal family line
• Paternal line. Recommendations from the user’s paternal family line
• Tree hot spot. Recommendations near an active area of the user’s tree.
  

As part of the split test, the team decided to randomize the order to better understand the value of each filter.

Secondly, we tested a variant that simplified and streamlined how tree information appeared at the top of the page. 

Before & After 

Slide the divider horizontally to view the improvements

Results

Over a series of weeks, the team split tested two variants against a control. Once the metrics reached statistical significance, they revealed:

• After a long drought of flat tests, this feature increased our primary metric, clicks per user, by a whopping eighteen percent
• Several secondary metrics that also measured clicking behaviors increased modestly
• Nearly all guardrail metrics were flat or positive
• The Maternal and Paternal filters yielded the most clicks out of all the new options we presented
• The variant that attempted to clean up the UI display yielded fewer clicks on several key widgets

After reviewing the overwhelmingly positive signals, the team released the most successful variant.

What I Learned

• Challenge the status quo with ideas that are rooted in a deep understanding of users. Behavioral data confirmed that users valued a convenient way to access the ancestors they had recently modified. However, during user interviews, many customers revealed that they often did not exclusively focus on one particular ancestor. Rather, users often concentrated on clusters of ancestors in certain branches or family lines. I leveraged these insights to add value to a feature that many regarded as a proven success.

Also find me @

© 2023 Kevan McGowan Design. All rights reserved