Overview

Michael Aram is a luxury home decor brand. With three brick-and-mortar stores, an eCommerce site, and a heavy presence at department stores like Neiman Marcus, Bloomingdale’s and Macy’s, they chose AdVenture Media to spearhead the overhaul of their advertising strategy on Google, Facebook, Instagram and Pinterest.

Goal

Michael Aram wants to:

• Increase brand awareness
• Capture a wider audience 
• Increase eCommerce sales  ‍

Problem‍

The Account Management team wanted to find out, for high-value customers, defined as repeat customers, which products they purchase for their subsequent orders. Is there an identifiable pattern that can help inform our advertising strategy?

Since the goal is to grow eCommerce sales by increasing brand awareness and capturing a wider audience, the Account Management team wanted to know which products serve as an entry point to Michael Aram. There are hundreds of products in their store. If there is an identifiable group of products serving as an entry point, these products can be the main feature to increase new customer acquisition. The subsequent group of products can serve in nurturing new customers and converting them into repeat customers.

Data Analysis

We exported two data sets from the client’s store. The first data set identified all the repeat customers. The second data set is a list of orders by date with other unique identifiers such as customer ID. With these two datasets, we filtered out any customers that made only a single order. The cleaned data set contains all orders made by repeat customers only.

With the cleaned data set, we used Python to iterate and group the products by 1st order, 2nd order, 3rd order and so forth. Python allows us to sort and group products by order sequence efficiently. The Python code starts with the first row and identifies the customer’s ID. It then referenced the whole data set and identified any other orders made by the same customer. Once it has identified all the orders made by this customer, it groups the products purchased into different lists using the order timestamp to sort them into the correct lists. Once this process is completed, it iterates downwards until all orders have been grouped into their respective lists.

With order sequence of products separated out by lists, we can now visualize these data sets and observe any identifiable product groups that served as an entry point for our high-value customers.

Results

We discovered that the entry point for a product did not have an impact on overall revenue from that customer. There was no observable group of products that served as an entry point, but rather, there was a group of products that drove most of the sales whether it was the 1st order, 2nd order and so forth. We cross-referenced this insight with a secondary analysis looking at all products, which included single-order customers. A group of products drove 80% of the revenue for the business. This product set was similar to the product set for repeat customers. We concluded that a group of products drove sales for both single-order and repeat customers. For repeat customers, they might buy product A and then come back to buy product B. Or they might start with product B then back to A.

As a result, our advertising efforts now focus on a group of shortlisted products. This analysis was instrumental to the creative team because we no longer have to test as many creatives. Instead, we can focus our efforts on making the already impactful creatives work and look better.