03 Launching Walmart’s 1st AI support chat
RoleTeam Lead | UX Design | Prompt Design
Opportunity
By enhancing Walmart’s existing automated support chatbot with GenAI, customers will get answers they need more quickly, with direct, natural language.
As GenAI gets more powerful, users will not need to go through multi-step flows to input information. Instead, AI will recognize the user’s intent, use Walmart’s data, and deliver a solution, ideally within a single cadence.
We believed AI could help solve customer problems more efficiently.
Our goal was to increase self-service and reduce escalations to agents.
ProcessThe design team worked with engineering, legal, and product to define a transparent and helpful experience for users. We provided a training prompt to Engineering to define voice, tone, and personality. We also initiated a continued audit of the source content that trains the AI FAQ responses.
Once core personality and capabilities were defined, my team
designed a mulit-variant A/B test to understand how customers preferred to interact with the AI versus existing preset flows. Through A/B testing, we learned there was no need for an explicit entry into an “AI” experience. Customers just wanted answers, they didn’t want to have to decide how to get them.
Outcome The team launched a pilot AI program in 2023, with a plan to ramp to 100% AI supported experience in 2024.
Throughout the pilot iterations, we have been able to double the self-service rate of AI-driven conversations and expect improvements to continue.
Live today
This same query may have taken 4+ clicks for the user to find their answer previously.