This case study is password protected
Don't have it? Reach out at yl.yuxuanli@gmail.com
Incorrect password. Try again.
← Back to workAs QuickBooks evolved into an AI + human platform, I worked with cross-functional teams to define how expert services could become part of the core product experience, including where support should appear, what kinds of help should exist, and how those services should work across AI and human touchpoints.
A big part of this work was figuring out where expert help would feel useful inside QuickBooks. By mapping customer journeys and looking across key product moments, I helped the team identify where support could create real value and narrow the opportunity into 3 clear service areas: onboarding, bookkeeping review, tax estimates.
For each service, I worked across product, ops, and expert teams to define 3 services from 0 to 1. That meant shaping what kind of help customers needed, when it should appear, and how AI and human support should work together, ensuring a seamless customer experience rather than one that felt awkwardly tacked on.
We launched MVP expert-assisted onboarding in Jan 2026, followed by bookkeeping review in February 2026, which gave us early signals across different product moments.
A 19% click-through rate on bookkeeping review and a +3.2pt lift in onboarding engagement both pointed to the same thing: customers were open to expert help when it showed up in the right moment, tied to a real need. Those signals helped the team sharpen the model and gave us confidence to expand expert help into more moments.
Design can create the most leverage when it helps teams make decisions before the product is fully defined.
I learned that in ambiguous spaces, design can have the most impact by helping teams define the right opportunity, shape the right service model, and use early signals to make better product decisions.