Walmart Connect’s Smart Recommendations
Using data science models to create personalized recommendations for campaign performance.
Background
The Problem
Timeframe: Sep 2023 – Aug 2024
Role: Lead Product Designer
Platform: Web
Collaborators: 4 Product Managers (across separate workstreams), 5 Frontend and Backend Engineers, Principal Product Designer (from sister team, Display Self-Serve), UX Researcher, UX Copywriter
Walmart Connect Ad Center is an internal platform for first- and third-party sellers to create both online and in-store ad campaigns, helping them reach the right customers and grow their businesses. Its core offering, Sponsored Search, enables advertisers to feature selected products and brand videos as search results to relevant customer search queries.
I enjoyed the challenge of working alongside this talented, cross-functional team of designers and collaborators, navigating a complex organizational structure where product, engineering, and product marketing (our key business stakeholder) each operated under separate leadership.
Sponsored Search was initially primarily catered to power users – experienced, ad-tech-savvy advertisers who often compared and contrasted multiple platforms, and even built their own tools to optimize performance. But with Walmart’s push to expand into the self-serve third-party seller market, we anticipated an influx of amateur advertisers who would need more guidance to effectively run campaigns and justify their ad spend.
Through user interviews and usage data, we got to know the two clearly distinct user groups: those who knew exactly what they were doing, and those who didn’t. They differed in their familiarity with advertising, the amount of time they could dedicate to campaign management, and the level of support they needed to feel confident completing key tasks.
However, they aligned on some foundational truths:
All were highly conscious of their ad spend and expected returns, with a want to minimize risk of loss
None fully trusted Walmart’s recommendations due to a lack of transparency
Most wanted a simpler, more automated way to drive results
So the challenge became: How do we design a system that supports beginners without alienating experienced advertisers? My goal was twofold: create an experience that builds trust and clarity for novice users, while maintaining efficiency, control, and depth for advanced ones.
The Opportunity
The Design
At the time, our platform lacked intuitive and actionable recommendations to support users. Existing suggestions not only lacked transparency and were hard to trust, but they also suffered from usability and scalability issues.
Meanwhile, our data science team was in the process of developing new algorithms to generate personalized recommendations for budgets, bidding strategies, and target metric goals. It was the perfect opportunity to reassess the end-to-end user journey and identify where these insights could have the most meaningful impact.
After extensive alignment with the design team and cross-functional stakeholders, we established a guiding principle: users should move from less detail to more detail as they navigate from the Homepage, to All Campaigns, to Campaign Editing.
Homepage serves as an entry point where users get a high-level snapshot of all their campaigns and orient themselves within the platform
All Campaigns allows users to compare performance across campaigns and view recommendations in context, helping them prioritize and allocate resources most effectively
Campaign Editing offers the deepest level of detail, where users can take action with full context and confidence
This progressive disclosure model ensured that users could start with a bird’s-eye view and dig into specifics as needed, surfacing the right level of information at the right time.