Mobile Grocery Recommendation System Using Predictive Algorithms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
New customers in retail stores using mobile devices take longer to complete their first shop due to limited customer data, resulting in lower conversion rates and reduced ability for retailers to customize the shopping experience, which hampers customer loyalty and retention.
Innovation Solution
A system that uses a mobile application to present products to customers, allowing them to add items to a grocery basket, with a data model and algorithm predicting likely purchases based on user data and cue card responses, utilizing recurrent neural networks like Tensorflow to enhance customer experience and learning over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If new customers use search and browse functions to shop on mobile devices, then they can find products, but the heavy interaction takes more time and reduces conversion rates
Solution Approach 1:
The system performs preliminary actions by proactively presenting product recommendations to customers before they need to search. The machine learning model predicts and displays products the customer is likely to want, eliminating the need for customers to initiate search actions and reducing interaction time.
Solution Approach 2:
The system serves itself by using its own collected data and machine learning algorithms to automatically generate and present product recommendations. The system learns from customer behavior patterns and autonomously curates personalized product displays without requiring customer effort to search or browse.
2Adaptability or versatility
If retailers collect more customer information upfront to customize experience, then personalization improves, but new customers are required to enter lots of information which reduces conversion
Solution Approach 1:
The system applies partial action by collecting only the minimum necessary information (postal code) initially and gradually gathering more customer data over time through usage patterns. This incremental approach provides basic personalization immediately while allowing the system to learn and improve customization as more data becomes available, without overwhelming new customers.
Solution Approach 2:
The system implements feedback loops where customer interactions with recommended products are continuously monitored and fed back into the machine learning model. This feedback mechanism allows the system to refine its understanding of customer preferences over time, improving personalization accuracy without requiring upfront information collection.
3Ease of operation
If first time shoppers use mobile devices with limited customer data, then accessibility improves, but conversion rate drops compared to desktop computer
Solution Approach 1:
The system performs preliminary actions by pre-calculating and displaying personalized product recommendations based on the customer's location data (postal code) and real-time behavior. This proactive approach compensates for the lack of extensive customer data by using machine learning to predict relevant products, maintaining high conversion rates on mobile devices.
Solution Approach 2:
The system dynamically changes parameters by adapting product recommendations based on real-time customer interactions, location data, and contextual information. The machine learning model adjusts product displays, pricing, and promotions on-the-fly to optimize conversion rates for mobile users with limited initial data.
Data Source
AI summary
A system and method includes presenting a single product to be located in the retail store to the customer using a mobile application running on the mobile device, receiving an indication from the customer either to dismiss the product or to add the product to a grocery basket of the mobile application to form user data, presenting a cue card to ask the customer a question, receiving an answer from the customer as to yes or no to the question to form cue card data, using a data model and an algorithm to predict the most probable product the customer is going to want to add to the grocery basket, transmitting a representation of the most probable product, and displaying the most probable product on a graphical display of the mobile device, wherein the most probable product is based on the user data and the cue card data.