Prevented over 19,250 agent contacts and increased engagement with, and discoverability of, support center content, leading to the ability to provide concise in-app help.

Starting in 2022, Amazon Ads decided to merge their sponsored ad platform and demand-side platform into one console. With the two separate platforms came two different support repositories. We knew through previous research that advertisers did not have easy access to help content that was relevant, effective, and accessible, making them less confident in their skills. Having this in mind, I saw an opportunity to not only unify this content but also provide a more thought-out learning and support experience.

With two different support centers came the problem of disjointed content structures across these repositories, even though they were owned by the same product team. To figure out how this was affecting advertisers, I started with a small research study of fifteen participants. Each participant was of varying levels of expertise when it came to advertising with Amazon Ads. What I found was that the majority of our advertisers valued different learning modalities, whether it be video, articles, or courses. They also had issues with the way our current content was structured.

Knowing that we had an opportunity to work on our content structure, I connected with our content team and ran a workshop to define how we would move forward with our content structure. In that workshop, we decided on a simple tree structure, starting with goals and objectives and moving into subcategories within those, then on to topics.

With the knowledge gained from our research and the completion of the content workshop, I started iterating on the homepage. When that was complete, I brought our designs to advertisers and received positive feedback around our goal / topic-based content structure. We also found that our advertisers wanted different learning modalities, like videos and courses. 

Some anecdotes from a lightning round of research on our updated homepage iterations.

The last bit of the puzzle to solve was our sub-pages. This included individual topic pages and content topic pages which housed the sub-topics of our goals. We landed on a mixed structure displaying sub-topics related to the goals and other learning modalities related to that goal as well.

With all the the updates to our structure and pages we needed to update how users interacted with search. In the original support center users had no type-ahead and a lackluster results page with zero filtering. To solve for this we added type-ahead, worked with our content team to revamp our metadata and provided easy and efficient filtering on the results page.

With this launch we prevented 19,250+ agent contacts annualized and increased engagement with and discoverability of support center content. We continue to see positive feedback on ease of access and our content structure!

If I were to work on this project again, I would think about how these patterns scale as we create more content and would potentially adjust the content structure to accommodate for this.