Enhancing E-commerce Experiences

How we used AI to transform customer experiences for
a leading U.S. Grocery Retailer

Our client is one of United States’ largest grocery retailers, serving more than 11 million customers daily in stores and online. With 2,800 stores in 35 states, our client operates two dozen grocery retail brands, 34 manufacturing locations, and 44 distribution facilities, totaling over $140 billion in annual revenue.

6%

Increase in #
of orders

0.3x

Increase in customer
loyalty

18+

Product categories
analyzed

0PB+

Customer and transaction data analyzed

Key
Takeaways

Processed terabytes of customer data and shopping behaviors to construct an AI recommendation model.
Developed distributed data pipelines on Azure Cloud for ETL, enrichment, and prediction.
Achieved a measurable ROI by boosting basket size through suggested products.
Created model training and tuning mechanisms using user-generated signals and feedback.

Challenge

Our client observed a common issue among customers: difficulty finding products matching their preferences. Consequently, customer interaction and interest decreased, resulting in reduced engagement, lower loyalty, and a decline in repeat business. This inadequate search experience led to customer frustration, dissatisfaction, and potential loss of sales. Additionally, they faced the challenge of fully grasping the vast amounts of data they possessed, like customer, transaction, and behavioral data, to generate better insights and enhance convenience for each customer across all channels.

Solution

Within a span of four weeks, we were able to design data cleaning operations as well as a machine learning engine that delivered real-time recommendations and personalized upsell and cross-sell suggestions to millions of customers utilizing the client's web or mobile apps. Each time a customer interacted with the client's products, it sent feedback to the engine, continuously improving the performance and accuracy of the recommendations and shopping experience.

Personalized Recommendations
After cleaning and analyzing petabytes of data, like order history and purchasing behavior, Rapidops employed advanced algorithms like collaborative and content-based filtering to develop a personalized recommendations engine that met customers’ unique preferences and needs.
Curated Product Listings
Rapidops utilized machine learning algorithms to track onsite behavior and real-time customer data points to personalize customer recommendations in more than 30+ categories. The engine curated product listings based on each customer’s preference and presented them in an order that aligned with their tastes and needs. It included suggestions like “Most Popular in this Category” and “Trending in this Category,” both of which led to increased conversions and repeat customers.
Dynamic Search Results
Rapidops implemented natural language processing, machine learning, and collaborative filtering to analyze customer search queries, browsing behavior, purchase history, and other relevant data points to generate tailored and relevant search outcomes in real-time. By eliminating the zero-result pages and humanizing the search experience with smart autocomplete, both customer engagement and the potential for successful conversions increased.
Inventory Predictions
To enhance customer experience and order fulfillment, Rapidops used machine learning to better predict customer baskets and improve inventory accuracy. Incorporating multiple data sources like pick events, reservations, velocity of products, seasonality, demand forecast, and supply chain trends enhanced predictions. Customers could then monitor instances of low-stock or out-of-stock products while placing orders, increasing order rates by over 5%.