Community Q&A
When review volume plateaus, a new contribution type becomes critical. I led the design of Community Q&A, a complementary contribution model that drives lightweight user-generated content for sustainable content growth.
Role
Lead Designer (Sole)
Team
1 PM · 6 Engineers · 1 Data Scientist
Year
2024–2026
I led design end-to-end as the sole designer on this project, working with a PM, 6 engineers, and a data scientist. I shaped the product strategy, conducted all research, and owned every design decision from early concept through launch.
01 — Context & Opportunity
Yelp's contribution model had long centered on reviews, but reviews couldn't answer everything. People had questions that cannot be easily answered by reviews, and there was no easy way to ask or get answers from the community. At the same time, review volume was plateauing, which made finding new ways for people to contribute more urgent.
We explored whether a community driven Q&A model could:
Open up new ways for people to contribute beyond writing reviews.
Lower the barrier to participation. Asking and answering a question is quicker and easier than writing a review.
Add a layer of knowledge that works alongside reviews, not in competition with them.
02 — Product Evolution Strategy
Community Q&A did not launch as a single feature. It developed in three stages: defining the vision, validating the user need, and then scaling the system.
Stage 1: Vision Work
Establish the right foundation before any design decisions were made.
Before building anything, the main challenge was understanding where Q&A fit within Yelp's existing experience. The risk was creating something that felt tacked on, overlapping with reviews or adding noise instead of value.
Key Design Questions
How should Q&A fit into Yelp?
How can it integrate without becoming a separate system?
Who participates?
Who asks questions, who answers them, and how do people move between reading and contributing?
How does intent become a question?
How does a broad intent, such as searching, turn into a specific question?
Design Foundation
Q&A as a complementary layer, extending search and reviews rather than competing with them.
Key surfaces across the journey
This phase established a shared direction for the product: Q&A as a complementary layer. It should extend search and reviews rather than competing with them.
Which user flows to focus on?
A key goal of the vision work was to surface the gaps between what Yelp currently offered and what users actually needed. The asking journey illustrates this most clearly. It's where unmet intent is most visible. I focused on two asking flows in the vision: search and AI chat. Here I'll use search as the example, since it represents the highest volume entry point for unanswered questions.
Stage 2: Pilot / PMF Test
Validate whether Q&A felt natural within Yelp without forcing the behavior.
With a direction set, the next question was: would people actually use this? Each milestone was structured as a hypothesis test that only justified the next stage if it proved out. M1 seeded questions from Elites to test if asking behavior existed at all (it did: 2.17% CTR, above the 2% benchmark for other SERP components). M2 used LLM-generated questions to test answering behavior without needing organic question volume first: ~1,000 answers in 20 days, with 40% of answerers giving more than one answer. Only once both sides of the exchange proved out did we invest in the full live Q&A system.
Pilot Focus
Establish the interaction model
Consistent structure for questions and replies, supporting both asking and answering.
Cover key surfaces
Identify discovery and contribution entry points within existing product flows.
Design Foundation
Prioritized consistency over customization, keeping Q&A close to Yelp's existing interaction patterns so it felt familiar, not foreign.
Key screens launched in Pilot test
Question asking
Prompt user to ask questions on SERP
Question answering
Solicit answers on Home and Post Review Screen
Design Outcome
- Established a clear, consistent interaction model for asking and answering
- Validated that the structure held up across both contribution types
- Created a stable foundation to build on in Stage 3
Stage 3: Scale
Make the system grow without breaking down.
Once we confirmed genuine demand, the challenge shifted from proving the concept to ensuring the system could grow coherently across surfaces.
Friction Reduction
- Converting search queries into questions
- Streamlining the reply experience
- Reducing cognitive load throughout the flow
Channel Expansion
- Expanding Q&A to Web and Android
- Introducing email and push notifications
- Adding more entry points to improve discoverability
Ecosystem Building
- Lightweight reactions to close the loop
- Establishing the Community Q&A Hub as a central destination
03 — Selected Design Initiatives from Stage 3
These initiatives span multiple layers of product design, from shifting user behavior and integrating systems, to refining contribution quality and feedback mechanics.
Growth
Convert Search Query to Community Question
Many searches on Yelp were actually questions, but there was high friction for users to turn that intent into a community interaction.
Design Challenges
- Shift the experience from transactional search to conversational asking
- Reduce the effort of composing a complete question
- Balance getting more contributions with keeping content quality high
- Align across Search, AI, and Ranking teams who all had a stake in this surface
Strategic Design Decisions
- Keep question creation lightweight and in context, so it didn't feel like a detour
- Use progressive prompting only when the user's intent was unclear, to avoid interrupting confident searches
- Make the human element visible: real answers from real people, distinct from search results or AI chat
Key Design Moves
- Built a mechanism that recognized when a search looked like a question and offered to convert it
- Designed a lightweight question creation flow to minimize the effort of asking
- Aligned the experience with how people think about search, so the transition felt natural
- Shifted the experience from passively browsing results to actively starting a conversation
Impact
- Increased both the rate and quality of questions submitted
- Turned a passive search behavior into active community participation
- Created a reusable pattern for intent-to-contribution conversion that could scale across other surfaces
Systems Building
Community Q&A Hub
As Q&A expanded, the content became scattered with no single place to browse, revisit, or engage.
Design Challenges
- How can Q&A integrate into the existing experience without becoming a separate system?
Strategic Design Decisions
- Who asks questions, who answers them, and how do people move between reading and contributing?
Key Design Moves
- Built a dedicated hub where users could browse, revisit, and engage with all Q&A content in one place
- Added Q&A hub entry points across key surfaces: Home, Yelp Assistant, Business pages, and more
- Organized questions geo-based to surface locally relevant content
Impact
- Increased Q&A visibility across the app, which drove higher engagement and answer rates
- Improved retention by giving users a reliable place to return to their questions and activity
- Laid the structural foundation for a scalable community ecosystem
Interaction Craft
Business Tagging & Prompting
Questions and replies often had no clear connection to specific businesses, making content harder to navigate.
Design Challenges
- Help users add business in their replies without feeling constrained or directed
- Balance AI driven suggestions with user autonomy: the system should assist, not dictate
- Introduce prompting in a way that works for all users, without overwhelming those who don't need it
Strategic Design Decisions
- Show the prompt upfront to prioritize education and drive feature adoption
- Trigger tagging based on signals in the user's reply: helpful when relevant, invisible when not
- Prioritize interaction quality over feature complexity, keeping the flow smooth and easy to follow
Key Design Moves
- Dynamic prompt suggestions
- Inline clarification cues
- Structured composition flow with progressive guidance
Key Interaction Decisions
Should users be able to change location while asking?
Removed manual location editing. Location is inferred automatically.
What is the right interaction for business tagging?
Used inline tagging to keep the flow focused and avoid breaking the user's train of thought.
When should business chips appear?
Show business chips only before the user has tagged a business, only needed as a starting point.
Impact
- Improved the quality and relevance of replies by grounding them in specific businesses
- Strengthened the connection between Q&A content and the businesses being discussed
- Improved the reader experience by making it easier to navigate directly to a relevant business page
Feedback Mechanism
“Love” a Reply
A lightweight way to close the feedback loop for contributors.
Design Tradeoff: Reaction Placement
Option A · Expandable Interaction Row
- Built with future growth in mind
- Supports future interaction expansion
- Feels sparse in the current state with reactions only
Option B · Inline Placement (Final)
- Optimized for the current experience
- Keeps the layout compact and natural
- Avoids introducing empty or unused space
Impact
- Made it easy for readers to show appreciation, closing the feedback loop for contributors
- Added a lightweight interaction that increased engagement without requiring extra effort
- Reinforced participation by giving contributors a signal that their answers were valued
04 — Outcome & Impact
Community Q&A launched as a new content type on Yelp, and the results showed it was filling a real gap.
- Q&A volume reached 10% of Yelp's review volume by Q1 2026, demonstrating that the community was willing to participate in a new way beyond writing reviews.
- The project unlocked a new contribution behavior at scale. Asking and answering questions became a meaningful activity for users who had never written a review before.
- Several interaction patterns developed for Q&A (including progressive tagging, inline prompting, and the question thread design) were added to Yelp's design system, giving future teams a reusable foundation to build on.
Reflections
Building Q&A from scratch and seeing it reach scale taught me a lot about what it takes to introduce a new behavior to an established product.
- The pilot phase was the most important investment. Shipping something small and focused early gave us the evidence we needed to build with confidence. It's easy to skip validation when you believe in the idea. The pilot is what turned belief into a credible direction.
- Designing for an existing ecosystem requires restraint. The temptation to invent new patterns is real, but fitting Q&A into Yelp's existing language made adoption faster and the experience more coherent. Novelty isn't always the right goal.
- Scale surfaces decisions you can't anticipate early. Many of the Stage 3 challenges (fragmentation, cross-surface consistency, contribution quality) only became visible once the product was real and growing. Staying close to the product post-launch was as important as the initial design work.