How a data visualization tool helped me develop a service design strategy
I attended a meetup named Storytelling in UX with my colleague recently. One of the presenters, Janet Taylor talked about her vision of streamlining their research strategy at Perforce and coming up with domain-wide user personas instead of having them specific to the products. They are still working on the final solution at Perforce, but her talk inspired me to share the similar process we went through for a new Accela solution earlier this year.
In this blog post, I am going to talk about Spider Graph — a data visualization tool we used in our design strategy process. The post is divided into two parts,
- The first part talks about the use of the tool to define the strategy.
- The later part talks about the data visualization process we went through to make the strategy more transparent to the bigger audience.
The Strategy
After interviewing 15 potential users, visiting 3 agencies and surveying 75 people for the new Accela solution — Recreation and Resource Management, we started synthesizing the bucket full of data. As we were focusing on holistic user journey that includes multiple touchpoints as a part of service instead of looking at a particular product or medium, we decided to define domain-wide personas. People are more than the type of activity they like, so we ended up with multiple (to be exact, 6) user characteristics and motivations as listed below.
- Proximity (Visitor, Local)
- Time (Holidays, Routine)
- Nature (Planning, Spontaneous)
- Interests (Location, Activity)
- Approach (Exploration, Adventure)
- Involvement (Environment, People)
We wanted to map the individual user data with the behavior spectrums in a visual way to see patterns (I am sure consuming it in text form above proves my point). Having 6 spectrums made it a challenge to visualize this data in a consumable way. We started exploring 2 options, the first one was to create multiple 2x2 matrices with different combinations of spectrums. In that case, we would end up with 36 matrices to compare with. While the other option was to create a multiple-dimension matrix. From the research, “Spider Graph” or “Radar Chart” seemed like a most viable tool to represent this data considering it can handle multiple axes. We decided to try Spider Graph as an experimental tool and iterate with time.
Mapping each user’s data using this tool helped us see common behavior patterns between different users. Below is an example of a grouping of the users based on similar data patterns.
We ended up with 4 different groups with similar pattern graphs. Each user group was given a persona archetype based on the representative user behaviors. We aggregated each group’s data to define one spider graph per group, and these final graphs became our persona graphs.
Below is the iterated version of the spider graphs for the 4 key personas for the Recreation and Resource Management domain. Having a very distinct spider graph for each persona helps us quickly understand each persona behavior in a visual way.
The Data Visualization
We went through multiple iterations to reach to the visualization shared in image 2. Image 3 is the first iteration of Spider Graph created by plotting each parameter data on each line. Although there are parameter pairs that define the spectrum, the data is plotted independently to get the full perspective.
We got feedback that it was difficult to match the relevant parameter pairs — i.e. it was not clear if Activity was compared with Location or Planning. At the same time, it was difficult to figure out how the data points are distributed along the lines.
To address the previously mentioned concerns, following 4 versions in image 4 were designed. Middle circles are introduced to give a clear idea of the data distribution along the lines. Colors and connecting lines are being used to clearly define the spectrum pairs. Version 4 — the bottom right one was the definite winner based on the feedback.
In the next iteration showed in image 5, each spectrum is defined with the parameter initials to give more clarity to the readers. A critique for the variations of the latest iteration gave us following feedback,
- Although the spectrum lines are highlighted and defined better in this version, we still need a clear distinction as it’s easy to compare the wrong parameters with each other.
- Having the full parameter names are more helpful while reading the data vs. having a list with the parameter initials.
We also explored an option of a simple two-dimensional ‘Data Matrix’ as an alternative to Spider Graph as shown in image 6. Although, the information from Data Matrix is easily consumable, the spectrum parameters are not independently mapped. As information accuracy is equally important along with visual simplicity, we decided to work on one more iteration of Spider Graph to make it more visually consumable instead of going in the direction of two-dimensional Data Matrix.
Image 7 is the final iteration of the Spider Graph we have created so far.
Conclusion
In this process, we learned that the Spider Graph is not an intuitive and quick learning tool, but it helps define patterns and present complex data in a visual way. They became the foundation for our personas, story maps and other parts of design strategy.
If you are working with multidimensional data where multiple recreational activities or different flavors of coffee define your user type, this tool can be very useful in plotting the data in relation to each other and find the patterns between different users and characteristics. I would love to hear your thoughts about different use cases for this tool as well as different tools for the same use case described here.
