This past summer, Andrew Gray served as a UX Developer Intern here at Ipreo. He shared the following story about his experience, and a specific project he worked on, with the Ipreo Tech Blog.
After years of uncertainty about my career path, something clicked during my internship and I knew that UX was my calling.
I wanted to share one of the many projects I was able to work on during my time at Ipreo during which the combination of the technical yet creative design process sparked my excitement and helped solidify my professional direction.
I was given the opportunity to revise and update a visual dashboard in order to provide more meaningful data analytics to internal business users. This was all done using Splunk which is a software that allows you to search, monitor, and analyze big data. My hopes are that this dashboard will provide quick and useful information to product managers and their teams.
My Design Process
Persona: Meet David
Although this dashboard has several types of personas, I am focusing on David, who would like some reassurance on how his product is doing. David is a hard-working Certified Scrum Product Owner (CSPO) by day and a caring family man by night. David spends his time developing and maintaining a clear vision for his product while making sure to make it home for his son’s baseball games.
Heuristic Evaluation of Key Pages
The layout of the dashboard in question was buggy and disorganized, and I was tasked with finding ways to improve it. I used a few usability heuristics from Nielsen Norman Group’s 10 Usability Heuristics to analyze and identify the problems with this UI.
Understanding the User
I then decided to take a step back to better understand the needs of the customer. This involved interviewing users in order to shape each panel around their needs and wants. I discovered that the majority of users have little to no experience with Splunk, its query language, or general programming. This reveals two core needs of the user:
- Need someone to create the dashboards – lack experience
- Need the dashboard to be quick to use – time-starved
This stresses the gravity of having a pre-made dashboard that’s efficient and intuitive so that no time is wasted on unnecessary Google searches or help pages.
Ideate Solutions Through Product/Market Fit
Out with the old, and in with the new. I modified the original panels and split them into more detailed, easier to read components while adding a few suggestions from users.
Card sorting is a method used to help design and assess the information architecture of your product. It’s a great way for users to categorize your products components in a way that makes sense to them. Four users went through this process and helped create a more enjoyable user flow.
I learned that quick information like # of users, # of customers, # of logins, etc. were placed more toward the top of the page while more detailed data like top pages and top referrers were closer to the bottom.
Wire-framing is a great technique that allows you to iterate quickly with little cost to time or energy. Based on the trials above I came up with a low fidelity mock-up which improves the dashboards structure and functionality.
Redesigning the dashboard through a higher fidelity helps to communicate form and function better. It’s also a huge time saver when you can accurately measure and validate the design before actually building the product.
The actual implementation of the panels was more complicated than what I originally thought. When I was given this assignment I had used Splunk approximately zero times. So I started with what I thought were the easy panels like # of clients/# of users and eventually (through trial and error) picked up on the query language. The filters were my last task and they actually took me the longest. This was because every panel that I had created now needed an updated query to accommodate the latest filter.
I demoed the resulting prototypes to a few new users. I created two slightly different dashboards and conducted A/B testing to gauge the users’ response and get some real world validation. The feedback I received reached consensus around one of the two options, and I had a winner.
After several iterations of the prototype, I finally found one that works. But through use of the dashboard I am certain that updates will be needed and additional panels will be added. So I think it’s safe to say that this will be equity_dashboard_v2.1 and not equity_dashboard_final_v2_final_really_final. Such is the life of a developer!
Thankfully, the dashboard I created has been a huge success with both the product and development sides. This comes from the ability to match user needs to what’s actually on the screen. Plain and simple. (But not so simple).
When I was assigned the task of building and updating this analytics dashboard, I assumed that it would have little to no effect on the team that I was working on, let alone the company itself.
I was ecstatic to see Ipreo prove me wrong.