Self-Service Advertising
Amazon Marketing Services (AMS) is Amazon's self-service advertising platform.
I started by driving an analysis of all existing tools and domains to get all stakeholders to a common understanding.
Working with cross-functional stakeholders, I mapped out the various ingress points and lifecycle of our platform vendors.
I designed a unified campaign management vision that handled registration, consideration, building, and launching campaigns.
"There are so many types of ads on Amazon. I feel like I'm not using the most effective ones, but I have no way of knowing for sure."
Working with stakeholders I created three high level campaign categories to help advertisers understand campaign types by visual recognition
Wireframing keyword bidding
An example of some quick wireframes I created to help vendors and sellers target keywords with specific bid amounts
Prototyping and testing
I designed and prototyped a self-service ad builder, scaling our campaign creation by putting templates into the hands of non-creatives. With this rich interactive prototype, I was able to test and refine the design before handing it off to development.
Prototyping and testing. I designed and prototyped a fully functional keywork power prediction tool, to ensure that it met our advertisers' needs and use case. I even simulated backend delay in showing the power ratings for each word. Developing this feature was a careful balance of sharing valuable targeting information with the vendor but not disclosing Amazon's proprietary numbers to competitors who might use the system. This feature was single-handedly responsible for boosting our volume of paid keyword clicks by more than 2X. And it helped tailor more relevant ads to our shoppers. A true win-win for the vendors and our customers.
I designed and prototyped this campaign dashboard, using existing vendor campaigns, interviews, and usability testing to introduce powerful features, such as intelligent campaign suggestions.
"These new dashboards and flows are a game changer from our current methodology."
