My role:
Worked with:
Overview:
I first began working on Code Artifact in January of 2018 as an internal artifact repository tool for Amazon developers. We always knew we’d launch it as an AWS product in the short-term, so the foundational work that was done for internal developers was intended to grow into a console experience. I continued iterating on the console experience into October 2019 when I was reallocated to a higher-priority project, but I continued to mentor the front-end engineer who took over the UX and development after I transitioned.
Process:
I began by sitting down to interview the PM and lead front-end and back-end engineers. We drew diagrams of what we thought the process should be together. Then I began to work on mockups based on our design system. Within a month, I did the first round of usability testing, and I discovered that there were two main concepts that were confusing to users, domains and repositories. The domains were necessary because of the way that AWS handles accounts and identifies but they were challenging to explain to customers. The term repositories was confusing to many users who were unfamiliar with competing products, like Artiifactory. They generally thought of the term repository as a source repository and struggled to understand why they’d use an artifact repository, especially because many of them used artifacts directly from providers, such as npm, maven, PyPI.
After the study, I took the findings to the team and made a number of recommendations. I made a number of changes to the UI and began advocating to remove the domain resource type and tried doing several more usability studies with various terms for repository. When I realized how big of a problem this was, I escalated all the way up to the director level. Yet, after many months of discussions about how to handle the domain process, the team decided to proceed with the existing resources, despite the data I had gathered, so I did my best to find terms and experiences that were easy for customers to understand and added some diagrams and helper text / clues throughout the experience to help customers understand the model. In October 2019, the leadership team decided to reallocate me to work on the next largest project, which was taking priority over the Code Artifact launch. The UX team was short-handed, so the front-end engineer took over for me. We continued to have regular meetings and I provided a lot of mentorship and oversight into the final designs.
When the product launched in June 2020, we had done 5 usability studies with a combination of internal and external customers, and made a variety of usability improvements based on our findings. I influenced a team that included the Sr. SDM, SDM, and 6 SDEs and FEEs to make significant changes to the system architecture and designs based on usability findings.
Outcome:
While I can’t expose the actual number of weekly actives we have, at the time of writing this post (3-months post launch) the number of actives for Code Artifact is +0.3% above what was forecasted for this time period. Even though it is still early, the service has seen a roughly 4% WoW growth rate, and about 48% of customers are net new to the AWS Code Suite. Since launch we’ve added CloudFormation support and are working to add support for additional languages along with other highly-requested features. We’ve also already hit several of our high-level goals, including acquiring at least 10 customers who download at least 100k asset downloads per week over a 4-week period. There are other exciting things happening with this service in the months to come, which I can’t yet disclose.
Prototype:
