Data Monetization: Advanced Audience Search
Background
User feedback consistently showed that finding the right audience on our platform was frustrating and inefficient. The root cause was a combination of complex data structure and an outdated search experience. As our data offering expanded, the problem became more severe instead of better.
At the same time, we saw an opportunity to leverage AI to simplify discovery and help users focus on intent rather than system structure.
Team
1 Product Designer
2 Product Manager
1 UX Researcher
Timeline
Feb 2024 - Aug 2024
My Role
- Led the vision exploration, focusing on system reframing, rapid prototyping, and early validation with users and stakeholders
- Shaped how AI could be integrated into the exploration experience by defining intent based interactions
What’s wrong with the existing experience?
Current Audience Exploration Page: Grid View
- Broken Information Structure
The current hierarchy forced users to navigate through too many layers.
If a user try to find BMW purchasers in the past three months had to dig through six levels just to find a relevant group and be exactly clear about how data are structured, which is insane. This is not how users think.
- Weak Search Capability
Search only supported exact keyword matching.
If a user searched for luxury cars, they would not see segments like BMW buyers or high income car shoppers unless the exact keyword appeared in the name. This disconnected user intent from system output.
“I know what I want but I don’t know what to type”
— Internal User
How we explore design opportunities?
To ground the project in real user needs, we conducted 12 in depth interviews (6 internal users and 6 external users). The interviews helped us uncover where users struggled in the current exploration process and what they considered an ideal way to discover and evaluate audiences. These insights gave us a clear problem frame and concrete opportunity areas for the design direction.
Building on this foundation, I led a cross functional brainstorming workshop with 2 designers and 3 PMs. We explored possible solutions across the full journey which included the starting point, discovery, evaluation, and targeting.
Problem Statement
❓
How might we help users explore audience more effectively and accurately?
Initial Direction
Shopping-style audience exploration
We started by studying several intuitive consumer exploration experiences such as shopping and media platforms, where users can browse large inventories without understanding any underlying structure. Inspired by this, our early direction was to create a landing experience where users could go straight to meaningful segments with AI guiding discovery, instead of navigating layers of data hierarchy.
Some mi-fi ideas during ideation and iteration
To understand what matters most to users, we ran 2 rounds of user testing with 10 internal users and rapidly iterated on the prototype. These sessions helped us refine the content strategy and interaction model.
* Note: Only the landing page is shown here. But the exploration covers the entire flow.
But what’s not enough?
This approach improved discoverability and made the experience more approachable, especially for new users.
However, our research showed that experienced users consistently began with search. When search failed, the entire exploration experience fell apart. Users also came with highly specific and evolving needs that browsing alone could never predict.
It’s not about browsing, it’s about search.
Refined Direction
Search-first, intent-driven discovery
As we gained a better understanding of user behavior, we began a new round of ideation. The earlier exploration still had its value, but testing showed that search needed to be the primary entry point. Our goal became finding the right balance.
Some ideas during ideation and iteration
After another round of user testing, we finally comes to the right direction that we want to go forward with.
What’s next?
This direction proved to be the right path. User feedback across the full flow was very positive and confirmed that a search first, intent driven model creates real value.
Before we can fully deliver this vision, we need to strengthen the foundational structure. I have been working closely with the PM to clean up the underlying data, flatten the hierarchy, improve audience naming rules, and enhance filter logic based on the new model. These foundational updates are already in development, which means we are now much closer to realizing the complete vision.
Final Vision
Natural language, AI powered exploration
The direction also proved highly scalable. When we first explored the future of audience discovery, our focus was primarily on machine learning and AI driven recommendations. However, as generative AI advanced rapidly, we began a new round of exploration using the same foundation. This time the work moved much faster because the redesigned structure and search first model were already in place.
The vision remains search first, but now powered by natural language interaction and more personalized results. Users can describe what they want in their own words, and the system interprets intent with far greater accuracy and flexibility.
Due to confidentiality, I cannot share the full set of final design details. But below is a high fidelity preview of the updated landing experience from my prototype.
