Cloud Optimization Experience
My Role: Strategy, User Research, UX/UI Design, Interaction Design
The Problem
CloudHealth by VMware is a leading Public Cloud Management Platform. Rightsizing is considered one of our top product features because of its highly beneficial cost savings and optimization. However, based on an analytic analysis we discovered that only 11% of users were actually utilizing rightsizing. Of the 11% that were gaining some benefits, all of those users were considered sophisticated Cloud Administrators - no new users. Most couldn’t see the value of the tool, found it “difficult to use” and “didn’t trust the data.”
​
How could we create a sticky product experience where new and advanced customers can get value from our optimization tool immediately without requiring extensive onboarding, training and customer support?
Project Goal
Improve rightsizing feature adoption to make our users successful at optimizing their cloud environment.
Research & Discovery Methods
-
Competitive Analysis
-
Customer Interviews
-
Personas
-
Cross Collaboration Workshops
-
End to End Customer Journey Map
-
Usability Testing
Research Approach
Before we could brainstorm solutions we needed to learn about our users needs, motivations and frustrations. I reached out to our top customers that were using the existing rightsizing tool, as well as our top customers that we thought would benefit from rightsizing, to answer a few key questions.
-
When it comes to optimization, what is your number one goal?
-
What is your current rightsizing process from start to finish?
-
What are your struggles and challenges?
-
If you’re not rightsizing today, why not?
-
Who is performing the activity of rightsizing at your company?
Research Insights
-
“Recommendations are valuable, but we don't trust the data from CloudHealth.”​
-
“I need a way to see that my teams are meeting our organization’s efficiency targets”
-
“Rightsizing in CH is extremely difficult to use. I’m the only one that uses the feature because training my teams would be too time consuming.”
-
“Policies are confusing and difficult to find/edit.”​
-
“Teams have targets they want to hit. I can’t easily see if they’re hitting those targets, and neither can they.”​
-
“I really can’t tell if we’re being efficient.”​
Customers just couldn’t figure it out. Rightsizing was an easy concept to understand, and we had made it so difficult to do.
Key Takeaway
Old Rightsizing Design
Personas
Based on these interviews as well as input from our internal CAM/TAM team, I identified two personas for this project. We referred to them throughout the entire product development process.
Primary Persona
Shauna, a CloudOps professional, is asked by her manager for her team’s infrastructure efficiency report. She signs into CloudHealth hoping to gather optimization metrics to present to her manager, however it’s not clear what the data means or if it’s reliable. She reads some documentation on CloudHealth Help Center, but she’s still not confident how to present it in a meaningful way to her manager.
“My manager is looking a summary that says we are operating at 80% efficiency. This will take me days to decipher and manually calculate.”
Design Studio Workshop
I organized a cross-functional design studio where I presented the problem & pains from customers, and we generated fresh ideas as a group. ​It’s a quick way to validate ideas and leave with a single cohesive design direction.
Through these activities, I was able to create a few important UX goals.
UX Goals
-
Simplify the overall experience and dramatically speed up the time to value.
-
Identify and showcase high level valuable content and communicate efficiency effectively.
-
Show the math - increase data integrity.
-
Streamline the workflow making it easy for users to complete their task in one place.
-
Increased valuable and customizable experience for advanced users.
-
Create a universal framework - same experience for all services, & all clouds.
Customer Journey Map
I developed an End To End Customer Journey Map to communicate customer pain points, visualize opportunities, and ways to track metrics along the way.
Sketches to Wireframes to High-Fidelity Prototype
Now it was time to take everything I learned from the research sessions to draft the initial sketches & wireframes, and rapidly iterate to arrive at the high-fidelity prototype. The goal was to quickly put a prototype in front of users and stakeholders and get their feedback.
Tools
Figma, Sketch, InVision
Prototypes Produced
Wireframe Prototype, Hi-Fidelity Prototype
Prototyping & Iterating
Workflows
Tooltips & Popovers
Recommendations
Efficiency Targets
Research Insights Drive Design Changes
I contacted all the original customers that I had spoke with to test the prototype and gather feedback.
Before Testing
After Testing
Research Insights
-
89% found the new design “easy to understand”
-
Efficiency score was an important metric to Cloud Admins, so I made that more prominent
-
RI% was valuable context necessary for making informed optimization decisions
-
Current vs Projected exec summaries were not obvious to the user
-
CloudOps wanted a way to view by team, so I added a grouping enhancement
-
Blue coloring for ‘over-utilized’ was confusing since it was a negative.
-
Status column with color variations tested to be distracting and also not a11y compliant
Usability Testing
I organized the first company-wide internal usability session for anyone to come and try out the new rightsizing design. I created a script, gathered my design team and trained them to conduct my usability test so that we could gather as much data as possible in a short timeframe.
Usability Insights
-
93% found the new design “easy to understand”
-
New metric chip quick filters were a usability win. 91% completed the task with no issues.
-
Creating a new target was a discoverability issue
-
No clear difference between current instance and recommended change
-
Metric colors in data-grid were not a11y compliant
-
Status labels and cost comparisons were not as clear as I thought
Research Leads to Improved Design Solutions
I consolidated the insights, presented the key take-aways to my team, as well as improved design solutions to reflect the feedback. We also received some enhancement feedback from our sessions. Using an impact effort matrix, I helped the team rank feature proposals.
Final Prototype
Simplified Optimization Experience
Feature adoption increased by 26% in just 90 days after Beta launch.
“This high-level summary is exactly what my director is looking for and saves me hours of time trying to figure out these number manually.”
“I know right away these recommendations are going to save me X amount with clear statement I understand.”
Custom Efficiency Targets
Old rightsizing policies were located in a different area of the platform, difficult to understand, and not influenced by the recommendations. Because of this, these policy blocks were extremely low utilized.
New Efficiency Targets keep the user on the Rightsizing page, are directly connected to recommendations, and are a huge usability improvement.
​
Creation of custom efficiency targets increased by 235% just 90 days after Beta launch.
New
Old
Improved Workflow
Time to value increased exponentially. Insightful default reports are available to users with zero effort on day one.
Users saved hours of time creating custom rightsizing reports.
Creating a custom report: Old workflow
25+
clicks
3
pages
6
hours
Creating a custom report: New workflow
<5
clicks
1
page
3
hours
Recommendation Choices
By displaying 3 rightsizing recommendations along with supporting data for each, customers had all the context necessary for making impactful decisions within CloudHealth.
Confidence in recommendations scored at 91% in our customer survey 90 days after Beta launch.
New Recommendations
Old Recommendation
Accessibility & Mobile Responsiveness
Numerous accessibility improvements were made across the entirety of the feature, and I created designs for mobile responsiveness for the project’s next phase.
Measuring Success
Qualitative and quantitative metrics were set up to measure the success of the project. Customer interviews were conducted and surveys were sent out 90 days after the feature launch.
-
Feature adoption increased by 26%
-
Creation of custom efficiency targets increased by 235%
-
Confidence in recommendations scored at 91%
-
Customers reportedly saved ~6 hours of time creating a highly valuable custom rightsizing report.
-
The new rightsizing design was successfully designed as a universal framework to fit all services. It’s now being implemented for other rightsizing services including: VMs, SQL DB, RDS, EBS, and Kubernetes
Lessons Learned
Incorporate a11y from the Start
During each research phase I uncovered more and more a11y issues with the design. If I had made accessibility a UX goal from the beginning, I would have had less iterations and therefore resulting in quicker time to value.
Persistence & Transparency
Getting a feature re-design on the product roadmap was tough. This project was de-prioritized a few times in order to make room for new features. It wasn't until we presented the improved experience and customer research to a wider audience that the project was fully funded.