Increasing Customer Engagement for a Q-Commerce Brand
About this project
A leading German q-commerce company aimed to implement a data-driven system to enhance customer engagement and elevate the activation rate within food delivery apps. Together with the client's experts, Netguru engineers constructed the Next Best Action mechanism, which analyzes data to select optimal activities to reduce customer churn.
Services
Industry
The client is a prominent e-commerce business with over two million customers, spanning more than 70 countries across four continents.
Their solution links customers with stores, supermarkets, and restaurants via dedicated apps and couriers. The platform pioneers q-commerce and caters to multiple verticals through a variety of brands.
The challenge - reducing churn for a global retailer
The goal of the project was to enhance profits by boosting customer engagement, and reducing churn in the client's apps. The client sought to interact with their customers in a way that would foster regular use of the platform.
Understanding how best to approach the customer was crucial. The client could then execute specific actions to engage customers:- send messages,
- dispatch emails,
- provide incentives such as vouchers or discounts.
It was critical to determine which actions would help persuade the customer, and which would be irritating and churn-inducing.
Netguru’s role and services provided
Solving the challenge required a machine learning pipeline capable of analyzing customer behavior and suggesting the next best action.
The client turned to Netguru for assistance in integrating the machine learning algorithm with the company’s existing infrastructure.
The client's internal data scientists developed the model. Netguru engineered the software and managed machine learning operations, such as developing infrastructure for predictions, creating tables in data warehouses, and integrating with existing communication channels.
Netguru engineers collaborated with the client in a team extension model, supporting the client's team daily.
Results
- The client received a comprehensive Next Best Action (NBA) pipeline with data gathering, feature engineering and calculation, model prediction, reward calculation, and reporting on predictions.
- The alpha version was built in one quarter, a notably tight timeline for such an intricate product.
- The NBA enabled the integration of additional functionalities into one of the client's products.
- The client received an automated system that sends vouchers or communications through an existing API, powered by the NBA pipeline, with a choice of eight different vouchers and eight push notifications with specific messages.
- The system can be readily extended with new actions, for example, assigning a percentage discount for the next order of a specific line of products, or a dedicated voucher with a personalized message.
- The intelligent system continues learning the most suitable action to choose for a specific customer at a given time.
Netguru’s approach to the project
The client needed to seamlessly integrate the Next Best Action algorithm with the organization's infrastructure without causing disruptions to the deployed pipelines. The design aimed to leverage available data, scheduling and deployment functionalities, and the integration APIs exposed for client communication (CRM, incentive creation, push notifications).
The Next Best Action system could function as an agent that either explores or exploits. Technically, every action is a trial and error test to verify which actions yield the expected results.
The results, accumulated over time and analyzed, allow the algorithm to learn which actions to perform with different customers for the desired outcome. Over time, the system will continue improving accuracy in choosing the most effective actions and action sequences.
