All case studies
Experiments, research, and the work that helped a team do both well. Choose the question that interests you.
Experiments and personalization
Finding the right moment for a recommendation →
Recommendations shown after an add-to-bag action increased revenue by 9% in the published team result. Timing made the idea useful.
When fewer clicks asked a better question →
A compact product rail got fewer clicks. Shoppers used search and the product grid more instead, giving us a better question for the next study.
A quieter invitation to sign in →
A smaller invitation improved reported sign-in and sign-up rates and was dismissed less often than a modal. The tooltip was implemented in production.
Making an existing member benefit easier to notice →
A relevant prompt helped eligible members activate an available offer. The next question was whether online activation also influenced store purchases.
Turning a recommendation model into a testable program →
We broke a broad personalization question into seven planned comparisons, with placements, audiences, tracking, and analysis handoffs made explicit.
Teams, process, and measurement
Getting product answers closer to the source →
I created Ulta Beauty’s Brand Expert Program to connect customers directly with brand knowledge. The internal workaround became a PowerReviews pilot and eventually Brand Engage.
Making experimentation a way of working →
The program grew from 20 to more than 65 tests a year in the published team account. I led the work of making good ideas easier to prioritize, deliver, and learn from.
Helping research travel beyond the researcher →
UX and product colleagues began running studies more independently. Shared standards mattered, and so did having someone available when they got stuck.
Making the assumptions behind value visible →
A common calculator helped distinguish test-period value from annual projections. The useful part was making the assumptions visible.
Agreeing on what trustworthy data should look like →
A small tracking request became a shared plan for dependable measurement: define the events, agree on variance, and give each kind of validation an owner.
Finding the delivery problem inside the experiment →
A later overlay iteration exposed rendering problems, environment differences, and unclear component ownership. The retrospective gave us specific changes to make in the next brief.
Coursework
When shade matching became a bigger question →
Five exploratory interviews about buying unfamiliar beauty products changed the question I wanted to answer. A Cornell product management course project.