Problem:

Highly engaged Mercari users want to know when they gain followers, but these notifications run the risk of cluttering the notification center, competing with high-priority alerts like, “You have an offer expiring in 2 hours”. How could we introduce notifications without overwhelming the user or our messaging system? 

Strategy:

The goal was to implement an aggregated notification, a best practice under-utilized in our ecosystem. Instead of sending individual alerts for each new follower, we create a single, comprehensive notification that is: actionable, informative, and less intrusive. Ultimately, I was working to keep the user experience clean while minimizing data usage on our side.

Results: 

The aggregated notification significantly outperformed the original, and set the groundwork for future campaign optimizations. 

My Story:

When the product team approached me with this notification request, the challenge was twofold: 

1. Improve community engagement through regular follower updates

2. Reduce noise for users who might not care about small-scale follower activity

The original follower notification had been stagnant since its launch in 2018—no updates, no A/B testing, nothing. Inspired by Meta apps like Instagram and Facebook, I proposed an aggregated approach: a single notification summarizing multiple follower events within a set timeframe. 

The technical challenge: how do we efficiently collect and deliver this data for millions of users without overloading our system? 

My product specification: 

• Segmentation: We needed a new custom event in order to narrow the scope of our audience to “has received a follow event in the last 24 hours”. This reduced unnecessary data processing across our multi-million user base. 

• Data Handling: Instead of temporarily adding follow data to user profiles for a single use-case, I requested a new Braze Connected Content endpoint to pull exactly what we needed—number of new followers and basic follower details via a single API call.

• Message Content: Using Liquid Logic, we dynamically populated the notification to scale from a single new follower to as much as hundreds at once. The message was always actionable: directing users either to the new follower’s profile to encourage a first message, or to their full following list. 

• Scheduling: To prevent inbox clutter, we sent one aggregated notification per user per day, in the morning, maximizing engagement while minimizing marketplace noise. 

Outcomes: 

As mentioned above, the aggregated notification significantly outperformed the original. Users were discussing it in forums, showing that the change provided a noticeable difference with real, actionable value. For the company, it reduced unnecessary data processing and simplified messaging management, a benefit for both the users and engineers. 

How would I improve this if the sky was the limit?

If I had an expanded engineering team, I would further optimize the notification center with AI: 

1. Categorized notification pills: users could filter messages by type, solving the lack of preference settings. 

2. Personalized engagement ranking: AI models could learn which notifications each user interacts with most, sorting notifications by relevance instead of recency. 

3. Smarter aggregated sends: For less active users, AI could generate a single daily push summarizing the most relevant updates across the site, increasing engagement while reducing noise. 

This approach would deliver a cleaner, smarter, and more personalized notification experience, improving both user satisfaction and platform efficiency.