Xtract
Xtract is a production‑level experimentation framework uniting design, strategy, and analytics. I led UX design by designing experiment variants, and partnering cross‑functionally to analyze results. Insights from live experiments were translated into scalable UX improvements across high‑traffic digital surfaces, ensuring design decisions were both creative and data‑driven.
It supports rapid testing across strategy, creative, web implementation, operations, and reporting. In 2022 alone, the pod contributed ~$25M in incremental LTV, accounting for 2% of online sales growth.
Below are a selected Xtract projects:
Global Sign-in Test
Primary metric: Increase sign-in CTR
Challenge: Users struggled to find the sign-in for the global navigation, creating friction and missed opportunities. The challenge was to make sign‑in easier to find and navigate, while keeping the overall navigation clean and consistent.
Approach: While the initial ask was to explore different sign‑in treatments, I pushed the idea further by introducing a delay‑triggered modal designed to gently prompt users to sign in. This approach aimed to align with user intent without disrupting the primary navigation experience.
Outcome: We ran an A/B/C experiment on att.com to test new homepage sign‑in entry points. Variant 2 emerged as the winner, increasing sign‑in link CTR by +6.22%, progression to sign‑in by +3.51%, and authenticated logins by +2.30%. A key learning was that prominent sign‑in placement significantly improves wayfinding—the below‑nav sign‑in link alone accounted for 7% of total sign‑ins for the winning variant.
Story Stack vs. Filmstrip Component
Primary metric: Click‑through rate (CTR)
Challenge: The homepage and key upper‑funnel pages needed a component that could communicate value quickly and clearly at page load. Existing layouts either required scrolling or split attention across multiple tiles, making it harder for customers to understand Signature benefits and take action.
Hypothesis: If the Signature Filmstrip component on the wireless plans page were replaced with a Story Stack, then click‑through would increase because key information would be visible immediately, without requiring users to scroll.
Key findings: Story Stack delivered higher click‑through by presenting content on page load, reducing the need for scrolling. Customers were more likely to engage with targeted areas of the site when using Story Stack
Learnings: Content that is immediately visible performs better than content requiring interaction to reveal. Component choice influences funnel quality. While Story Stack increased engagement, follow‑up analysis showed opportunities to improve downstream progression once customers reached Signature pages
Impact: Established Story Stack as the preferred component when introducing merchandising offers
B2B CKAV Lead Capture Testing
Primary metric: Engagement rates with the CKAV inline form
Challenge: The existing B2B CKAV experience focused on eligibility results but missed an opportunity to capture customer intent when interest was highest. At the same time, asking for contact information too early risked increasing bounce and breaking flow.
The challenge was to test if, when, and how to request customer information during CKAV in a way that felt appropriate for business customers.
Hypothesis: If we introduce a lead‑capture step during CKAV on the Business Fiber page, customers will be willing to share contact information, creating actionable B2B sales leads while maintaining engagement with the availability flow.
Approach: The variants tested were simplified form, form populating as user fills, and longer form
Key learnings: Business customers are more open to sharing contact details when the value exchange is clear. Simpler forms reduce perceived friction at high‑intent moments. CKAV is a strategic opportunity to capture demand—not just display eligibility
Impact: This work showed that CKAV can support B2B lead capture without hurting the overall experience. It helped clarify how much information to ask for, when to ask for it, and how to build trust through consent and messaging. As a result, design and strategy teams gained a tested, repeatable pattern for balancing conversion performance with lead generation goals.
Takeaway: Designing lead capture into a high‑intent B2B moment requires trading convenience for clarity—but when the value exchange is explicit, customers are willing to share information and the business gains incremental growth without sacrificing core conversion.
Snapshot of Xtract work
