Improving customer experience by making in-store purchases obsolete
Covid pandemic scenario changed the retail status quo and the e-commerce channel witnessed an increase, in a short period, that was only expected to happen in the following years. Consequently, the online channel is, now more than ever, a challenging area where retailers are hard-pressed to strive for innovation and excellence.
Our client, a large retail chain, had long devoted considerable effort to create innovative services and improve the customer experience. Two key challenges needed to be addressed:
- Which innovative service can be offered to the customer to face the pandemic?
- How to improve customer online experience?
The proposed innovative service aims to improve customer experience by offering weekly home deliveries with a suggested basket of products.
This service’s main goal is to eliminate the need for customers going to the store to buy regular products and streamline the process of building the e-commerce basket.
The first step was to build a machine learning model that predicts the probability of a family of products being bought each week, complemented by a heuristic — a set of intelligent rules — that chooses the products inside each family.
The model was built on drivers gathered from customer segmentation and transactional data. This data allowed us to build variables related to customer consumption behavior, product consumption profile, purchase history for each group customer/product, and seasonality.
Subsequently, an interface was developed to communicate the suggested basket to the online store. The customer just needs to confirm the suggested products, or add new ones, and close the order. Customer life gets easier, right?
The designed solution meets our client’s ambition to be at the forefront of innovation. Thanks to the effort of teams from multiple areas, working together with the same goal, the solution was deployed to gather feedback from a selected range of customers.
The most recent results show that 55% of our recommendations at the family level were accepted by the customer.
Inside each family, 60% of the chosen products were accepted with a 60 % accuracy in the quantity suggested. The entire solution was well received by the initial customers, with a NPS score ranging from 7 to 8.
Going forward, there is still room for improvement and the increasing amount of feedback will be crucial to take the predictive and prescriptive models to the next level.