Better IA through Card Sorting

Uncovered users’ mental models through card sorting exercises to determine an optimal information architecture for the #1 Wedding Planning App.

 

Type of Project

We ran into this IA challenge as part of a larger redesign effort to position the app as the indispensable comprehensive wedding planning tool.

My responsibilities

Information Architecture Ideation, User Validation, Planning & Scope Definition, Stakeholder Alignment, Leadership Buy-In

Time

Aug - Sep 2017

 

The challenge:

As part of this massive re-design effort, we are tasked to combine features from two separate apps into one. On top of that, there are two additional brand new planning features coming down the pipeline, so the IA of the mobile app needs to accommodate all the changes without overwhelming the end users.

 
 

Building Alignment:

Based on the data insights and information hierarchy of existing IA, we were able to quickly draw the line in the sand to separate the essential menu items from the rest. The good news was that internal stakeholders and senior leadership reached the agreement on the “essential” items within a meeting; the bad news was we still had 12 “essential” items left in the pile. “Do we have to keep the junk drawer moving forward?” The PM asked me reluctantly. “Let’s try to fit all of them within 5 tabs. Simple is hard but it’s usually better! Otherwise, more junk will appear in the junk drawer in no time.” I responded with determination. Knowing card sorting would be the right methodology to help inform the new IA, we started jotting down the list of participants that we would like to invite. In addition, I outlined the following criteria to  help the group evaluate the potential mobile navigation IA together. 

Coherent, Minimal, Consistent 
The deeper a hierarchy becomes, the more likely visitors are to become disoriented.
— Nielsen Norman Group
 

Paper Card Sorting First: 

9 stakeholders from cross-functional teams participated in the card sorting exercise. We prepared the 12 menu items on sticky notes ahead of time, and asked each participant to bucket those items based on relevancy. The only requirement was to keep the total number of buckets within 5. I was optimistic that we would be able to identify trends and reach consensus internally, so we could move onto user validation next. Unexpectedly, 9 people generated 6 distinct IA proposals. There were some insightful discussions and debates, but the outcome made me less confident to make a final IA recommendation for the app. 

 
 

Stuck, Now what? 

Coming out of the card sorting exercise with stakeholders, we knew we needed to take a different approach to reach alignment. There was a lot of chatter around the feature labels. In addition, a number of moving pieces were undergoing re-design effort which increased the uncertainty and confusion of this IA exercise for some participants. After going through my notes, I realized labeling should have been part of the Card Sorting exercise. In order to level the playing field, instead of using feature names as cards, we decided to examine the 12 essential items and write down all the main jobs to be done (JTBD) associated with them.   

 

Let the Users Chime In: 

For the second round, we conducted a digital Card Sorting exercise with a much wider group of audience. Big kudos to the Data Science/Taxonomy team for introducing the tool Optimal Workshop to our design team a few months earlier. The setup was very intuitive, we recruited existing users from the web via a Qualaroo pop-up and got results within 5 days. 

Card Sorting Round II…

154 The Knot users participated

Two independent studies (open and hybrid)

We asked users to group 19 tasks into several buckets and label them

 

What We Learned:

1.

Sure Bets

“Planning” is the most popular suggestions for the suite of tools  - Timeline + Checklist + Budgeter

“Inspiration” is strongly favored as standalone section of the app - Inspiration + Vision Quiz + Favorite Images

Vendors = Vendor Search + Saved Vendors + Reviews + Vendor Chat 

Guest List Manager makes sense to be its own tab, “Guest List” is the recommended label

This was the most insightful graph from the Card Sorting exercise, we leveraged it for many follow-up experiments during the re-design!

This was the most insightful graph from the Card Sorting exercise, we leveraged it for many follow-up experiments during the re-design!

 

2.

Serendipitous Learning

There is strong natural affinity between WWS and Couple Photo & Profile (+ 60%)

There is some affinity between Inspiration and Vendors (~50%)

 

3.

Areas to pause and reconsider

Correlation between Wedding Vision and Couple Profile is not as strong as predicted

Registry doesn’t have strong correlation with either Wedding Website or Guest List Manager

The Dentrogram along with Similarity Matrix provided a strong mental model based on tasks!

The Dentrogram along with Similarity Matrix provided a strong mental model based on tasks!

 
 

Results & Next Steps:

After gathering the user input on our tab bar navigation, we had a much more clear direction to move towards. Obviously, we didn’t just implement 100% of what the users had suggested because there were intricate business implications and multiple feature integration timeline we had to consider. I presented our multi-phased mobile IA recommendation along with the data findings, it was refreshing to see squad leads, designers and senior leadership nodding along in agreement. We received unanimous buy-in :-) Soon after, I started collaborating with the UI designer to create updated icons and interaction patterns.

Fast forward to 3 months later, the release of the All-In-One Wedding Planner from The Knot was a huge success. 63% growth of new members YoY, 74% increase of users interacting with multiple tools. This IA project was instrumental in setting the foundation to make the essential features more accessible and more useful. Read more about the re-design in my next case study!