FEATURE RELEASE
Stop and Shop: Search Typeahead
Stop and Shop: Search Typeahead
Stop and Shop: Search Typeahead
COMPANY
Stop and Shop
ROLE
UI/UX Designer
Impact
Add to cart Sessions (ATC): Increased from ~14% to ~20%
ATC Revenue: Grew from $800k to $1.4M
Add to cart Sessions (ATC): Increased from ~14% to ~20%
ATC Revenue: Grew from $800k to $1.4M
Add to cart Sessions (ATC): Increased from ~14% to ~20%
ATC Revenue: Grew from $800k to $1.4M
Overview
Overview
Overview
As the UI/UX Designer at Peapod Digital Labs, for the client, Stop and Shop, an online grocery platform, I led a cross-functional initiative to overhaul the platform’s search functionality. I partnered closely with engineering and product stakeholders to identify key pain points. Our goal was to address long-standing pain points and deliver a seamless, high-performing search experience. Through user research, competitor benchmarking, and close collaboration with design and engineering, we shipped a search redesign that significantly improved user satisfaction and drove measurable revenue growth
As the UI Designer and UX Researcher for the Stop and Shop online grocery platform, I led a cross-functional initiative to overhaul the platform’s search functionality. I partnered closely with engineering and product stakeholders to identify key pain points. Our goal was to address long-standing pain points and deliver a seamless, high-performing search experience. Through user research, competitor benchmarking, and close collaboration with design and engineering, we shipped a search redesign that significantly improved user satisfaction and drove measurable revenue growth
As the UI/UX Designer at Peapod Digital Labs, for the client, Stop and Shop, an online grocery platform, I led a cross-functional initiative to overhaul the platform’s search functionality. I partnered closely with engineering and product stakeholders to identify key pain points. Our goal was to address long-standing pain points and deliver a seamless, high-performing search experience. Through user research, competitor benchmarking, and close collaboration with design and engineering, we shipped a search redesign that significantly improved user satisfaction and drove measurable revenue growth
Why were we doing this?
Why were we doing this?
Why were we doing this?
- Increase in Bounce Rate: There was an exponential increase in Bounce Rate on Engagement Metrics for Search Functionality
- Limited Product Discovery: The existing search returned a narrow set of results, impacting both customer satisfaction and conversion.
- Low Relevance: Users frequently encountered results that didn’t match their intent, leading to friction and drop-offs.
- Desktop Underperformance: Desktop usage remained strong, but the experience lacked sophisticated navigation tools that could improve engagement and retention
- Limited Product Discovery: The existing search returned a narrow set of results, impacting both customer satisfaction and conversion.
- Low Relevance: Users frequently encountered results that didn’t match their intent, leading to friction and drop-offs.
- Desktop Underperformance: Desktop usage remained strong, but the experience lacked sophisticated navigation tools that could improve engagement and retention.
- Increase in Bounce Rate: There was an exponential increase in Bounce Rate on Engagement Metrics for Search Functionality
- Limited Product Discovery: The existing search returned a narrow set of results, impacting both customer satisfaction and conversion.
- Low Relevance: Users frequently encountered results that didn’t match their intent, leading to friction and drop-offs.
- Desktop Underperformance: Desktop usage remained strong, but the experience lacked sophisticated navigation tools that could improve engagement and retention
Project Timeline
Goals
Project Timeline
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Understanding Traction For Search Functionality
Understanding Traction For Search Functionality
Understanding Traction For Search Functionality
From some high level Quantitative Analysis we gathered the data that, 69% of shoppers use the search bar immediately upon arriving at an e-commerce site
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Defining Stakeholders
Defining Stakeholders
Defining Stakeholders
Our Stakeholders of Shop and Stop were Shoppers, Sellers, Vendors and Brand Owners. Out of which our target users for the feature change were Shoppers. We prioritized them based on impact and needs. Shoppers made the largest user base contributing directly to engagement and revenue. Also, they had high friction points in their user journey
Entities who own product brands and may use the platform for direct-to-consumer (D2C) distribution or brand visibility — distinct from sellers if they control product IP and manufacturing
Brand Owners
Companies who supply products in bulk to the platform or to sellers. This includes manufacturers, wholesalers, and distributors — typically B2B relationships
Vendors
Third-party businesses or individuals who list products for sale on the platform but do not manufacture the products themselves (e.g., resellers, local stores)
Sellers
Shoppers
Individuals who use the platform primarily to buy groceries for personal or household consumption
Entities who own product brands and may use the platform for direct-to-consumer (D2C) distribution or brand visibility — distinct from sellers if they control product IP and manufacturing
Brand Owners
Companies who supply products in bulk to the platform or to sellers. This includes manufacturers, wholesalers, and distributors — typically B2B relationships
Vendors
Third-party businesses or individuals who list products for sale on the platform but do not manufacture the products themselves (e.g., resellers, local stores)
Sellers
Shoppers
Individuals who use the platform primarily to buy groceries for personal or household consumption
Entities who own product brands and may use the platform for direct-to-consumer (D2C) distribution or brand visibility — distinct from sellers if they control product IP and manufacturing
Brand Owners
Companies who supply products in bulk to the platform or to sellers. This includes manufacturers, wholesalers, and distributors — typically B2B relationships
Vendors
Third-party businesses or individuals who list products for sale on the platform but do not manufacture the products themselves (e.g., resellers, local stores)
Sellers
Shoppers
Individuals who use the platform primarily to buy groceries for personal or household consumption
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
User Journey
User Journey
User Journey
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
User Analysis
User Analysis
User Analysis
We did a detailed survey of our users through feedback loops and analyzing user comments
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
SWOT Analysis
SWOT Analysis
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Insights
Insights
Insights
We gathered datapoints through User Journey, User Feedback and SWOT Analysis as well as other quantitative and qualitative analysis to set our goals. Our major insights were:
🔷 Broken & Inaccurate Search Undermines User Trust
📌 Users frequently encounter zero results for relevant queries (e.g., “yogurt,” “bagels”) due to poor synonym recognition and lack of typo tolerance.
🔷 Search Experience Lacks Personalization & Context
📌 Static suggestions and disconnect between user terms vs. site terms lead to drop-offs. There's a need for dynamic, personalized, and behavior-aware search suggestions
🔷 Search is Central to the Entire Purchase Journey
📌 Search is the entry point for discovery and impacts all downstream stages: evaluation, carting, payment, and retention. Failures here compound across the funnel
We gathered datapoints through User Journey, User Feedback and SWOT Analysis as well as other quantitative and qualitative analysis to set our goals. Our major insights were:
🔷 Broken & Inaccurate Search Undermines User Trust
📌 Users frequently encounter zero results for relevant queries (e.g., “yogurt,” “bagels”) due to poor synonym recognition and lack of typo tolerance.
🔷
🔷
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Goals
Goals
Goals
To drive both user value and business outcomes, we established the following key objectives:
Enhance Search Functionality
Deliver a smarter, more intuitive search engine that could better understand and predict user intent—ultimately improving conversion rates.Improve Browsing Experience
Create a more cohesive flow from search to product discovery by minimizing dead ends and optimizing the UI for decision-making.Increase Product Findability
Use real-time suggestions, filters, and machine learning to elevate relevant results and surface complementary items—supporting both UX and upsell potential.Modernize Layout and Information Architecture
Transition from a rigid tile-based display to a clean, efficient list view optimized for scan-ability, comparison, and accessibility.
Each goal was tied to key metrics, including reduction in bounce rates, increased session duration, higher search-to-cart ratios, and ultimately, revenue per user.
- Enhance Search Functionality
Deliver a smarter, more intuitive search engine that could better understand and predict user intent—ultimately improving conversion rates. - Improve Browsing Experience
Create a more cohesive flow from search to product discovery by minimizing dead ends and optimizing the UI for decision-making. - Increase Product Findability
Use real-time suggestions, filters, and machine learning to elevate relevant results and surface complementary items—supporting both UX and upsell potential. - Modernize Layout and Information Architecture
Transition from a rigid tile-based display to a clean, efficient list view optimized for scan-ability, comparison, and accessibility.
- Enhance Search Functionality
Deliver a smarter, more intuitive search engine that could better understand and predict user intent—ultimately improving conversion rates. - Improve Browsing Experience
Create a more cohesive flow from search to product discovery by minimizing dead ends and optimizing the UI for decision-making. - Increase Product Findability
Use real-time suggestions, filters, and machine learning to elevate relevant results and surface complementary items—supporting both UX and upsell potential. - Modernize Layout and Information Architecture
Transition from a rigid tile-based display to a clean, efficient list view optimized for scan-ability, comparison, and accessibility.
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
What Does Success Look Like?
What Does Success Look Like?
What Does Success Look Like?
ATC From Type Ahead
Engagement and measure TAU and unique ATC from the search typeahead
ATC From Type Ahead
Engagement and measure TAU and unique ATC from the search typeahead
ATC From Type Ahead
Engagement and measure TAU and unique ATC from the search typeahead
Decrease in Customer Comments
Through channels like, userVoice, App Store Reviews and Customer Care
Decrease in Customer Comments
Through channels like, userVoice, App Store Reviews and Customer Care
Engagement
Unique and active users engaging with Typeahead, Profit per order from ATC on Typeahead
Engagement
Unique and active users engaging with Typeahead, Profit per order from ATC
on Typeahead
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Stakeholder Presentation: Proposed Features and Prioritization
Proposed features in front of stakeholders
Stakeholder Presentation: Proposed Features and Prioritization
From our competitive analysis and user research, we came up with the proposal of the below features:
We prioritized based on Impact Vs Effort
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Integrating ML Powered Search UX
Integrating ML Powered Search UX
✨ Why ML was necessary
Traditional rule-based search couldn’t keep up with evolving user behavior, ambiguous queries, and personalization needs. Machine learning enabled the system to understand intent, correct typos, and learn from user interactions — turning search from a static lookup tool into a dynamic, self-optimizing discovery engine
We picked models that gave the most bang for our buck — the ones that made search smarter without slowing things down or costing too much to maintain
Key Features:
✨ Why ML was necessary
Traditional rule-based search couldn’t keep up with evolving user behavior, ambiguous queries, and personalization needs. Machine learning enabled the system to understand intent, correct typos, and learn from user interactions — turning search from a static lookup tool into a dynamic, self-optimizing discovery engine
We picked models that gave the most bang for our buck — the ones that made search smarter without slowing things down or costing too much to maintain
Key Features:
✨ Why ML was necessary
Traditional rule-based search couldn’t keep up with evolving user behavior, ambiguous queries, and personalization needs. Machine learning enabled the system to understand intent, correct typos, and learn from user interactions — turning search from a static lookup tool into a dynamic, self-optimizing discovery engine
We picked models that gave the most bang for our buck — the ones that made search smarter without slowing things down or costing too much to maintain
Key Features:
Models: Contextual Bandits (e.g., Epsilon-Greedy)
Purpose: Real-time Learning
Why Selected:
Provides real-time learning from user feedback (clicks, skips).Trade-offs:
Doesn’t model multi-step interactions
Needs careful exploration vs. exploitation balance
Requires robust logging & reward tuning
Impact:
Before ML: Same search suggestions shown to everyone regardless of behavior
After ML: Learns which suggestions work best for each user in real-time
Recommendation System: Hybrid Recommendations
(Collaborative Filtering + Content-Based Filtering)
Models: Matrix Factorization, Word2Vec
Purpose: Personalized Recommendations
Why Selected:
To combine user interaction patterns (Collaborative Filtering) with semantic product similarity (Word2Vec) for more relevant and diverse recommendations.Trade-offs:
Requires regular retraining to stay relevant
Cold start problem for new users or items
Balancing signal weight between CF and content-based requires tuning
Impact:
Before ML: All users saw generic popular items
After ML: Users received personalized suggestions based on behavior and product meaning (e.g., “wireless earbuds” after searching for “Bluetooth headphones”
Model: Levenshtein + Fuzzy Matching
Purpose: Typo Correction
Why Selected:
For typo correction — handles common misspellings like “iphne” → “iPhone”Trade-offs:
No contextual understanding
Limited to spelling variations
Can produce false positives without filtering
Impact:
Before ML: Typing “iphne” gave zero results
After ML: Auto-corrected to “iPhone” and suggested top products instantly
Model: Levenshtein + Fuzzy Matching
Purpose: Typo Correction
Why Selected:
For typo correction — handles common misspellings like “iphne” → “iPhone”Trade-offs:
No contextual understanding
Limited to spelling variations
Can produce false positives without filtering
Impact:
Before ML: Typing “iphne” gave zero results
After ML: Auto-corrected to “iPhone” and suggested top products instantly
Model: Sentence-BERT
Purpose: Semantic Matching
Why Selected:
Enables semantic matching, surfacing relevant products even if the exact words don’t match the query.Trade-offs:
Embeddings require offline preprocessing & storage
Slower real-time similarity lookup
Memory-intensive
Impact:
Before ML: Typing “home workout gear” only returned results containing that exact phrase
After ML: Matched related terms like “dumbbells,” “resistance bands,” and “yoga mats” using embeddings
Model: BERT / DistilBERT
Purpose: Intent Detection
Why Selected:
It understands the meaning behind ambiguous or multi-intent queries like “apple” or “running”Trade-offs:
High latency unless distilled
Requires GPU resources for real-time inference
Needs fine-tuning on domain-specific data
Impact:
Before ML: Typing “apple” showed a generic list of fruits and electronics mixed together
After ML: Understood context (e.g., past searches) → showed either iPhones or grocery items appropriately
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
The Solution ( Mobile)
The Solution ( Mobile)
The Solution ( Mobile)
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
The Solution ( Desktop )
The Solution ( Desktop )
The Solution ( Desktop )
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Data Analysis
User Journey
Data Analysis
To make the feature enhancement more effective, using my HCI knowledge, I decided to frame effectiveness as “how much does a user learn from using the product?” and “how do they react in unfamiliar problem situations?” I analyzed the user journey to see how much information the users can potentially retain and how well they react to new situations. I used this data to establish a benchmark for future iterations of the product
To make the feature enhancement more effective, using my HCI knowledge, I decided to frame effectiveness as “how much does a user learn from using the product?” and “how do they react in unfamiliar problem situations?” I analyzed the user journey to see how much information the users can potentially retain and how well they react to new situations. I used this data to establish a benchmark for future iterations of the product
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Outcome
The Stop & Shop Search Typeahead redesign delivered substantial business and user impact:
Increased Add-to-Cart Sessions
ATC sessions grew from 14% to 20%, indicating improved relevance and usability of search results.Boosted ATC Revenue
Revenue attributed to search-driven ATC rose from $800K to $1.4M, demonstrating a direct uplift in monetization through enhanced product discoverability.Significant Reduction in Customer Complaints
Complaints related to search functionality dropped by 75%, reflecting higher customer satisfaction and fewer friction points.Effective Cross-Functional Execution
As part of personal growth, I got to work closely within UX, engineering and analytics teams to deliver a data-informed, scalable solution—enhancing engagement and aligning with both user intent and business goals.
Outcome
Redesigned Stop & Shop’s Search Typeahead experience resulted in,
1) The ATC ( Add-to-cart ) sessions increased from 14% to 20%
1) ATC Revenue rose from $800k to $1.4M
2) 75% reduction in customer complaints.
3) Collaborated with cross-functional teams to deliver a data-driven solution that improved product discoverability and boosted user engagement across platforms.
Outcome
Redesigned Stop & Shop’s Search Typeahead experience resulted in,
1) The ATC ( Add-to-cart ) sessions increased from 14% to 20%
1) ATC Revenue rose from $800k to $1.4M
2) 75% reduction in customer complaints.
3) Collaborated with cross-functional teams to deliver a data-driven solution that improved product discoverability and boosted user engagement across platforms.
To ensure the redesign was grounded in user needs and industry benchmarks, the team
To ensure the redesign was grounded in user needs and industry benchmarks, the team
Final Thoughts
Search is deceptively complex. While it might appear to be a straightforward feature, it quickly reveals its depth when approached with the goal of delivering real user and business value. Early in the project, I underestimated its complexity—but through hands-on discovery, user research, and performance analysis, I came to appreciate just how nuanced user intent, behavior, and expectations around search can be.
One of the key learnings from this initiative was the importance of data-driven decision making. Users interact with search in diverse ways, and qualitative assumptions alone would not have uncovered the underlying friction points. By leveraging real user data, we were able to inform every iteration with confidence and precision.
This project also reinforced the critical role of cross-functional collaboration. Working closely with UX designers, engineers, and analysts in a small but focused team environment taught me that strong communication and alignment are not just helpful—they are essential for both delivering impactful outcomes and fostering a healthy, productive team dynamic.
The Stop & Shop Search Typeahead Redesign stands as a strong example of how thoughtful product management—grounded in user insights, aligned to business metrics, and executed through collaborative teamwork—can turn a core user experience into a strategic growth lever. The improvements we made not only enhanced the user journey across devices but also laid the groundwork for continued innovation in how customers discover products.
Final Thoughts
Search is deceptively complex. While it might appear to be a straightforward feature, it quickly reveals its depth when approached with the goal of delivering real user and business value. Early in the project, I underestimated its complexity—but through hands-on discovery, user research, and performance analysis, I came to appreciate just how nuanced user intent, behavior, and expectations around search can be.
One of the key learnings from this initiative was the importance of data-driven decision making. Users interact with search in diverse ways, and qualitative assumptions alone would not have uncovered the underlying friction points. By leveraging real user data, we were able to inform every iteration with confidence and precision.
This project also reinforced the critical role of cross-functional collaboration. Working closely with UX designers, engineers, and analysts in a small but focused team environment taught me that strong communication and alignment are not just helpful—they are essential for both delivering impactful outcomes and fostering a healthy, productive team dynamic.
The Stop & Shop Search Typeahead Redesign stands as a strong example of how thoughtful product management—grounded in user insights, aligned to business metrics, and executed through collaborative teamwork—can turn a core user experience into a strategic growth lever. The improvements we made not only enhanced the user journey across devices but also laid the groundwork for continued innovation in how customers discover products