Designing a Privacy-Preserving Phone Lead System

Executing on a high-impact, trust-sensitive roadmap

CHALLENGE

In early 2023, Yelp was just breaking even on its big bet for nascent market expansion. New consumers had not yet built trust and familiarity with the platform, and did not stay due to the high-friction flow. In order to improve lead quality, we needed to give consumers and service pros a faster way to connect. Enabling calls and texts were the clearest path — but doing so at scale required solving for privacy, trust, and operational complexity.

IMPACT

I executed design for a multi-milestone roadmap across consumer and pro surfaces, collaborating with Product, Legal, Marketing, Research and five engineering teams. The system launched globally in Jan 2024, resulting in:

• Contributions to 4× revenue from nascent markets (via pay-per-lead monetization),

• 75% vs. 49% project-level and 37% vs. 23% conversation-level consumer first-reply engagement lifts.

Beyond launch, I created a post-MVP roadmap anticipating future pain points, ensuring continuity and long-term success.

Context & Challenge

Prior to this work:

• Consumers only received batched SMS notifications and could not reply via text.

• Pros wanted to pitch live, but privacy concerns stalled past efforts around direct number sharing at Yelp.

Problem statement

How might we enable calling and texting across multiple flows and platforms, while respecting privacy and driving measurable engagement?

Approach & Execution

Multi-Milestone Roadmap

Under a design manager’s initial roadmap buy-in, I led design for all three milestones:

• M1: Real-time SMS notifications for consumers → +1.7% relative lift in consumer first-reply engagement.

• M2: Consumer replies via text.

• M3: Full two-way calling/texting with masked numbers across the pro and consumer experience, and consent across consumer surfaces.

Consumer-Side Design

• To start, I designed experiments to validate willingness to share phone numbers, which informed the initial roadmap.

• Scope-wise, the final experience spanned opt-in and communication flows for multiple entry points (originating leads, non-originating leads, search engine marketing leads), with security variations for different user types.

• Taking into account the engagement rate gap between nascent and mature markets, we optimized where we collected phone numbers and implemented OTP login to reduce friction - increasing login success in nascent markets by 28%.

Pro-Side Design

• Designed a scalable masked calling/texting foundation for the automated lead-matching pipeline product.

• Collaborated with product managers, three engineering teams and one additional designer who scaled this into the pay-per-lead product.

• Balanced privacy and usability trade-offs: masked numbers, transparent consent copy, and call routing/texting UX.

Designing for Integration & System Complexity

Process Notes

CREATING a shared VISION

With a clear roadmap in place, design and engineering kicked off in parallel. My first priority was creating a shared understanding of what a world with masking could look like across product, design, and engineering. To do this, I translated early technical explorations into a low-fidelity end-to-end flow that stakeholders could react to. This gave everyone — from execs to engineers — a concrete picture of the future state and allowed for informed feedback on each iteration.

taming complexity

To integrate masking with minimal disruption, I mapped the information architecture of the existing system: stack-ranking objects, identifying dependencies, and determining where masked numbers should live. This produced a blueprint of key screens and reusable patterns that guided my work. This was key for stakeholder reviews - I could quickly outfit components to compare design directions.

CLARITY WITHOUT OVERLOAD

A core design challenge was balancing clarity with transparency. Explaining too much risked overwhelming or confusing users; over-simplifying risked appearing opaque and eroding trust. I designed for different levels of context: consumers only needed to manage a single connection, while pros juggled many and required more control. This principle informed everything from copy to screen states.

advocating for the user under pressure

As the deadline loomed and scope-cut discussions began, I consistently acted as the voice of the user. For example, I successfully advocated for giving pros the ability to change their phone number within a masked connection to mitigate user error, demonstrating how it would scale for multi-location businesses. These conversations were possible because I had built deep trust with engineering: I included them in weekly design reviews, kept my solutions lean by using existing components, and showed how my IA approach made features scalable by design.

ADAPTING QUICKLY TO TECHNICAL CONSTRAINTS

The close partnership I built with engineering allowed us to handle unexpected challenges with confidence. Example: Midway through the project, we realized that relying on Twilio masked numbers at scale would be prohibitively expensive. We communicated early, and I quickly metabolized the technical phone number recycling system into designs.

delivering a scalable two-sided system

The result was a sleek, two-sided system integrated into a complex ecosystem without breaking workflows. It shipped on time, and I received positive feedback from design, product, and engineering leaders for the clarity and scalability of the solution.

Results & Impact

• Global launch: Jan 2024.

• Engagement lifts: Project-level first-reply 75% vs. 49%; Conversation-level 37% vs. 23%.

• Adoption: 63% of opted-in consumers actively using feature.

• Business impact: Contributed to 4× revenue contribution from nascent markets; 25% YoY services revenue growth