Create a way for developers to understand what is in their infrastructure across multiple resources: servers, containers, packages and hosts.
Puppet Enterprise was a leader in the management and automation of infrastructure. When I joined Puppet was looking to expand their product suite from one product to multiple. For 7 months a small scrum team had been working on a product that was trying to solve for a way to let users know what was in their infrastructure.
I came in with a new PM to help lead the team to create a tech preview release for PuppetConf 2017; with a fast general audience release following a month after.
(1) Lead designer
(1) Engineering scrum
(1) Lead PM
Interaction & visual design
Creating a vision and moving forward
For the past 7 months the team had spent time figuring out the backend technology for how to discover what was across a single infrastructure, but it still wasn’t clear to all of Puppet what the goal or first release would be for Puppet Discovery. I partned with the team to come up with an overarching storyboard for the product.
We continued to pick up from previous research work with users and think through the information architecture.
A default dashboard view could show every type of resource across the infrastructure but filtered views could either be based on resource type of overall application. We still needed to learn which was more important to the user and why.
There was also the interaction model for “clicking” on a single object within the dashboard versus objects within a data visualization to consider.
Defining and executing first release
However, PuppetConf was coming up quickly and we needed to determine what was in the initial release. We came up with the following “acts” for starting and scaling this new product from tech preview to general audience release.
The following “acts” focused the team on delivering the following user capabilities for tech preview:
Creating an account
Adding one or more sources
Viewing resources within your infrastructure
Completing a basic infrastructure task (like starting and stoping a server)
After months of hard work, I got up on stage at PuppetConf to share the hard work of the team.
We tested the tech preview live at the conference and monitored key tasks (i.e. connecting various sources) over the following weeks. Overall, we had an average of 85% success rate for key tasks with a ~74% rate for users perception of ease. Our conversion rate for task completion was 53.55%, so we had a lot to dig into to figure out why and where to iterate next.