Impact

- A $6.8M opportunity unlocked with efficient customer journey mapping

- Identified lost opportunity across the path to purchase, cross-validated data from various sources, mapped search information with corresponding categories

- Detailed customer cohorts created, conversion patterns tracked and monitored across stages and domains

Challenge

The client, a leading CPG company based in the EU wanted a deeper insight into customer purchase behavior. The company produced a wide variety of products ranging from dairy to plant-based and more, in order to cater to customers from different demographics. The client wanted to identify factors that define choices made by consumers and wished to identify the factors driving change in decision-making.

Approach

TheMathCompany worked with the client to leverage online customer sales data efficiently. In doing so, the client would be able to understand consumer preferences, identify pivotal points in the customer journey, and analyze purchase basket and customer purchase behavior. Furthermore, the insights would also help quantify lost opportunity, identify and segment cohorts based on customers’ demographic and behavioral attributes, and cull potential ways for increasing volume and value conversions.

Solution

To enable a streamlined customer journey, the data was cleaned using NLP techniques, following which customers were analyzed on various touchpoints across the customer journey funnel. Lost opportunity was assessed and subsequently customer profiles were created based on factors such as search homogeneity, product features and purchase patterns for each category. Additionally, potential ways to increase conversion on volume and value were identified.

Here’s a detailed overview of the solutioning process:

                                                                                                                                                                                                                                                                                                                                                                                                                    Fig 1: Overview of the customer journey funnel

                                                                                     Fig 2: Overview of the development of the customer journey cycle


Step 1: DATA PROCESSING

In this step, the customer search information was cleaned out and analyzed to determine customer behaviors and key channels. The product title, brand name, and product category/sub category information was also processed and classified.

                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      

                                                                                                  Fig 3: Overview of data preparation process


Step 2: FEATURE ENGINEERING

Brand flags were created to identify the client’s brand and competitor’s brand seamlessly. Product features were created to highlight key elements in the product that are influential in determining customer purchase decisions (e.g., products that are keto, superfood, plant-based, etc.). Then, the referrer domain was analyzed to bucket the users based on the way in which they accessed the target page, i.e., via emails, social media, or search.

Step 3: MARKET BASKET ANALYSIS & HYPOTHESES GENERATION 

The team gained a holistic understanding of the different products/brands that are being bought together, as well as the time of the day when the consumer purchased the product and the frequency at which the product was purchased.

Step 4: MODELLING

Categories of customers were determined via analysis of customer data using techniques such as text similarity and term frequency. Topic Modelling was used to synthesize search information, and technologies such as entity resolution modelling, random forest, and image classification were used to standardize product titles. Relevant algorithms and techniques were used to understand what products/brands were bought together and to determine customer cohorts.

                                                                                                  Fig 4: Overview of Customer Cohort Modelling


Step 5: MEASURING IMPACT

The team quantified lost opportunity in terms of value and volume across the purchase funnel for each category/brand/product for each domain. Then, the team explored the potential reasons for drop offs across the purchase funnel. These included the following:

• The switch to competitors/competitor products

• Affordability, i.e., the switch for value pack/discounts

• Availability, i.e., inventory/listing across domains

• Affinity, i.e., towards specific brands/products

• Attributes, i.e., missing product features

At this stage, the team was also able to identify whether customers moved to competitors.

Step 6: PRODUCT DEPLOYMENT

The solution developed using these extensive processes was deployed on a cloud-based platform to ensure that the data was refreshed daily.

Step 7: APPLICATION CREATION

A business front-end layer was set up, so that the end-users in the client company could easily utilize the customer journey solutioning and analyze the needed KPIs.

                                                                                               Fig 5: Screenshot of the Customer Journey Application


The path to purchase tool deployed to the client helped them with the following:

- Get a detailed insight into lost opportunities

The tool identified lost opportunity across the path to purchase by the mapping of brands, product categories, and product titles, and then cross-validating data from alternate data sources. Following this, the search, surf information were mapped with corresponding categories.

- Create customer cohorts

By using the search, product features and purchase patterns in the application, the tool identified similar consumers, developed consumer personas based on homogenous profiles, quantified intent based on profile & interest levels, and provided detailed insights into customers’ baskets in order to understand the likelihood of products being purchased together

- Gauge conversion potential

The tool mapped the customer cohorts across the purchase journey, tracked and monitored conversion patterns across stages and domains, and prioritized customer potential based on volume and value to increase conversion.

share case study

facebook linkdin twitter

Want critical data insights to unlock business value?

Get Started
mobilelogo
Linkedin Instagram Facebook Twitter
Cookies policy Privacy policy