Overview
Eggplant is the industry-leading test automation and monitoring SaaS platform that features deep learning and artificial intelligence to help test applications at scale.
"Lets rid the world of bad software"
My role
I was hired as the lead senior UX/UI designer to work across all of eggplants platforms to design a new SaaS application that combines 5 existing core services into 1 new product. Working closely with all stakeholders including product owners, lead techs and multiple development teams across the globe.
One challenge of many
The existing modeler designer sits in an old Dojo framework that was being updated and bought into the new React framework. The challenge was to reduce the adoption time of this feature which was causing customers using our trial to look elsewhere.
Discovery
Understand.
It’s important to note that the goal of product discovery is not necessary to ship features. Rather, it’s to promote an environment of learning that will help you improve your product incrementally and consistently. Discovery is an allocated period of time where you explore the problem space before any design or development takes place. Discoveries look very different from one project to the next in terms of the time it takes, who’s involved, and what the outputs are at the end.
​
Discovery is a great time to understand about, stakeholders, teams, goals, users, strategy, tasks, success metrics, and risks.
​
There’s no single, defined set of methods for product discovery — but broadly speaking, product teams follow four stages: learn & understand, define & decide and ideate & prioritize.
Product strategy:
I also went about finding out about what engineers' costs would be relative to the business goals. How the release would fit inside the roadmap. I worked with team leads and product managers and shared my product strategy to enable a release on time.
Research
I had already produced personas so I had these to hand when I was trying to empathize with the users. But I wanted to deeper understand the process of setting up a current model. I wanted to also look at Jobs-to-Be-Done: A Useful Tool to Focus on Outcomes Rather than Features.
I conducted numerous interviews, contextual inquiries, along with a focus group to gain real insights into the product we have. I was also trying to find out not how users were building models in our platform but what they were actually trying to achieve. I obtained lots of insights. I split this up into two. Feedback on existing features and new value proposition.
I also created user journey maps to get a better understanding of the flow. I started to put all learnings into consumable data and insights.
Research Analysis - Identifying the issues
What does the data mean for the product and user I am designing for? I identified patterns and common themes that I turned into concrete insights. I then presented these to all stakeholders. Along with many insights into controls and the process of building the current UI. I concluded that Eggplant and its users are disjointed in their vision for graphical models. Eggplant has been using the State/Action building method for a particularly long time and thought its users understood it. But users were always trying to convert this into language and visuals that they understand and use commonly.
"I want to see what I am building so I have 2 screens open at once"
"We refer to states as screens here"
"It would be easier to see what I am designing"
from the insights, I worked on a meaningful solution. The modeler would enable real application visualisation into the building process. Looking at the jobs to be done approach we found users wanted a real representation of their model, not a model representation.
Ideate.
User flows helped engage engineering and check on feasibility. As this was only a step in the process of running tests I wanted to always show it in the overall bigger picture and would always reference the user journey. Wireframes were created to gain usability knowledge under some quick testing.
Prototyping.
I built a full working prototype and begin running user tests using Useberry. I gained valuable data on areas and actions users were getting stuck on through heatmaps and video recordings. Some of the main concerns were as follows.
​
-
Action properties were still unclear and needed an order
-
Users needed to check the status of the state visual.
-
Users requested help to show a debug/pre flight run
The solution
Its worth noting I had already built a design system as seen below that was used in my prototyping.
The Solution was a new modeler service that would allow the user to upload a real image of the application and work over it to build a model that can be used in our existing platform for AI testing. Working closely with engineering we located an open source platform (cytoscape) that could be easier adjusted to what we wanted. Prior to agreeing to use it, we underwent intensive analysis to check compatailty. The design also was a new panel control system which I reworked to my findings in research. It now has a set process to each action and state that users before could not make sense of without extension knowledge of the product.
​
Development hand off went through the usual process with constent communication throughout. So how can we measure the success of the design, well first off we had instant approval from sales with the ability to sell in greatly improved. numbers to follow. The time on task for new users was decresed and our support for new users dropped. Analytical data is currently being gathhered. To design great products, you must deliver the right features, with the right user experience for the right people.
I am currently still at Eggplant delivering consistent results and success on this award-winning software.
