Improving users' online shopping experience

Online product recommendation system

The Client

Top Retailer

One of the top retailers in the USA with stores across all the 50 U.S. states, employing more than 400,000 team members, was looking for an online product recommender system for their e-commerce website.

Akaike’s edge-cloud agnostic solutions offer great flexibility.

Flexibility 0%
Cost 0%

Executive Summary

Industry Overview

Disruption

The global e-commerce market size was valued at USD 9.09 trillion in 2019 and is expected to grow at a compound annual growth rate (CAGR) of 14.7% from 2020 to 2027. Increasing penetration of the internet is bolstering the smartphone-using population across the world. Digital content, travel and leisure, financial services, and e-tailing among others constitute a variety of e-commerce options available to the internet-accessing customer base that is gaining momentum with increased internet usage. Therefore, technological awareness among customers is expected to deliver a positive impact on the market growth.
Business Challenge

Customer Experience

Right product recommendation is critical to customer experience and purchase decisions that help in generating better customer loyalty and higher revenue for e-commerce organizations. The customer wanted to showcase the right product recommendation to improve customers’ online shopping experience. In such a competitive e-commerce environment, they wanted to maximize the opportunity for the time spent by their visitors on the website in terms of experience and revenue generation. Therefore, an intelligent product recommender system is what the customer was looking for to increase the overall conversion rate as the desired and preferred products are shown to their new and returning users.

The Akaike Edge

Inbuilt libraries, DL models with transfer learning capabilities

Experienced ML and DL Ops teams

Efficient Deployment 0%
Integration 0%
Ongoing Maintenance 0%

Solution

Inbuilt libraries, DL models with transfer learning capabilities
Step 1.

Selection of the appropriate variant for Collaborative Filtering approaches

Collaborative Filtering (CF) is one of the machine learning techniques for recommender systems to cluster similar users or items together and can be done in different ways. Multiple iterations were conducted to finalize the most appropriate variant of the Collaborative Filtering type recommendation for the task to showcase similar products for the selected product by the user.

Step 2.

The solution was deployed in a distributed cluster

Based on the data, the CF recommender system was built. Tables were created for a set of products that were purchased and viewed, and put together for the Collaborative-Filtering process. The solution was deployed in a distributed cluster of 100 machines, providing recommendations for more than 250,000 SKUs/products.

Step 3.

A/B testing and product validation

After analyzing the incoming traffic of real-time users, we were able to demonstrate to the customer that our recommender system improved the click as well as the conversion rates compared to displaying a random product to the visitors under a specific category. Upon proper validation, all the online traffic was exposed to the CF recommender system, improving the overall conversion rate for the customer.

Research shows that 87% of Vision AI
projects do not yield expected results

either owing to training data insufficiencies stalling the project, or being too slow to deploy. Our AI experts can help you accelerate in data sparse environments.

We use cookies to give you the best experience.

Get in Touch