Topics

Context

Topics was predominatly a Marketing Tool aim at unlocking a new pathway for marketing and Business Intelligence teams to show Recommendations, sponsored content and native Advertizing.

All KPI's and hard experiment results not shared for confidentiality reasons.

Execution

Project: Mercari Follow Topics

The Mercari Follow Topics project was aimed at enhancing user personalization by allowing users to follow and unfollow specific "topics" related to items on the Mercari platform. This feature provides users with a curated feed of items associated with their followed topics, reducing the need for manual searches.

My Role

As a Technical lead of the Topics team, I was responsible for managing the topics and the components that make up the topic timeline. This involved defining and managing seasonal topics, sorting them by display period, and periodically collaborating with the Business Operations (BO) team to curate topics from various external sources like Business Analytics, Marketing, and even external companies. I also played a key role in the architecture and system design followed by the implementation of the topic-service microservice.

Tools and Technologies Used

The project utilized a microservices architecture, with the topic-service handling all aspects of topic management. We initially used an in-memory database for the MVP phase to quickly validate the feature with users. The service was designed to interact with other services such as the recommendation/search service and the item-details service. For the API endpoints, we implemented most relevant endpoints to fetch topic details and associated recommendeditems.

At the time of leaving the project, future plans included transitioning to a GCP Firestore NoSQL database for storing and serving data, and developing a React UI tool for more efficient management of topics by non-technical teams directly.

Results and Learnings

The introduction of the Follow Topics feature led to an improved user experience on the Mercari platform, with users being able to easily discover items of interest without extensive searching. The collaboration between the Topics team and the BO team was crucial in ensuring the relevance and quality of the topics presented to users.

One of the key learnings from this project was the importance of flexibility in the system design to accommodate future enhancements and changes. Additionally, the need for robust validation mechanisms became evident to prevent inappropriate data from being included in the topic feeds. Another significant learning was that for MVP delivery, delivering customer value and validating that value is more important than providing the best architecture and appropriate technology, which would take ages to take to the market. This approach allowed us to quickly iterate and improve based on user feedback.