Customer journey prediction and customer segmentation

Patent metadata

Field	Value
Publication number	US-11315132-B2
Application number	US-201916281374-A
Country	US
Kind code	B2
Filing date	Feb 21, 2019
Priority date	Feb 21, 2019
Publication date	Apr 26, 2022
Grant date	Apr 26, 2022

Abstract

Embodiments for implementing intelligent customer journey prediction and customer segmentation of a processor in a computing environment. A response outcome of a customer journey for a user may be predicted according to an assigned score based on one or more discriminatory sequence patterns identified between one or more groups of customers using one or more machine learning operations.

First claim

The invention claimed is:

1. A method for implementing intelligent customer journey prediction and customer segmentation in a computing environment by a processor, comprising:
receiving data representative of a customer journey for each of a plurality of customers by a computing device, wherein the customer journey is a plurality of actions sequentially undertaken by a respective one of the plurality of customers with respect to a potential purchase of goods or services of an entity;
training a neural network using the data to identify one or more discriminatory sequence patterns of those of the plurality of actions of the customer journey of the respective one of the plurality of customers that is similar to the plurality of actions sequentially taken by others of the plurality of customers, wherein the neural network groups those of the plurality of customers having similar customer journeys into one or more groups of customers using neural embeddings obtained from an event description of the plurality of actions, and wherein training the neural network includes:
performing a sequential pattern mining operation to determine similar discriminatory sequence patterns of the one or more discriminatory sequence patterns,
grouping those of the similar discriminatory patterns into the one or more groups of customers according to binary matches within the similar discriminatory patterns,
within each group, computing incremental coverage for each discriminatory subsequence of the one or more discriminatory sequence patterns, and
adding those of the discriminatory subsequences having the incremental coverage below a predefined threshold to one of the classes;

in conjunction with grouping those of the similar discriminatory sequence patterns into the one or more groups of customers, computing a lift of a signature of each of the one or more discriminatory sequence patterns as a ratio that the signature of a first one of the one or more discriminatory sequence patterns occurring in a first one of the classes also occurs in a second one of the one or more discriminatory sequence patterns in a second one of the classes, wherein the lift of the signature is used to determine when to compress the signature and when to discard the signature, and wherein timing information of the signature with respect to an event of interest is augmented to the signature such that the timing information is used to assign an importance to the signature as compared with an alternative signature notwithstanding whether the signature has a lower computed lift with respect to the alternative signature;
predicting a response outcome of the customer journey for a user according to an assigned score based on the one or more discriminatory sequence patterns identified between the one or more of the groups of customers using the trained neural network; and
displaying the predicted response outcome of the customer journey on a display associated with the computing device.

2. The method of claim 1 , further including identifying one or more friction points on of the customer journey.

3. The method of claim 1 , further including:
learning a successful response outcome of each customer journey generated from a first discriminatory sequence pattern associated with a first group of customers; or
learning an unsuccessful response outcome of each customer journey generated from a second discriminatory sequence pattern associated with a second group of customers.

4. The method of claim 1 , further including assigning one or more semantic tags associated with each sequence performed during the customer journey by the user, wherein the one or more semantic tags assist in identifying similar actions performed by both the user and the one or more groups of customers for identifying and learning the one or more discriminatory sequence patterns.

5. The method of claim 1 , further including grouping together the similar discriminatory sequence patterns according to one or more similarity metrics, a grouping strategy, or a combination thereof.

6. The method of claim 1 , further including assigning the score to the customer journey according to a number of sequences of the one or more discriminatory sequence patterns.

7. A system for implementing intelligent customer journey prediction and customer segmentation in a computing environment, comprising:
one or more computers with executable instructions that when executed cause the system to:
receive data representative of a customer journey for each of a plurality of customers by the one or more computers, wherein the customer journey is a plurality of actions sequentially undertaken by a respective one of the plurality of customers with respect to a potential purchase of goods or services of an entity;
train a neural network using the data to identify one or more discriminatory sequence patterns of those of the plurality of actions of the customer journey of the respective one of the plurality of customers that is similar to the plurality of actions sequentially taken by others of the plurality of customers, wherein the neural network groups those of the plurality of customers having similar customer journeys into one or more groups of customers using neural embeddings obtained from an event description of the plurality of actions, and wherein training the neural network includes:
performing a sequential pattern mining operation to determine similar discriminatory sequence patterns of the one or more discriminatory sequence patterns,
grouping those of the similar discriminatory patterns into the one or more groups of customers according to binary matches within the similar discriminatory patterns,
within each group, computing incremental coverage for each discriminatory subsequence of the one or more discriminatory sequence patterns, and
adding those of the discriminatory subsequences having the incremental coverage below a predefined threshold to one of the classes;

in conjunction with grouping those of the similar discriminatory sequence patterns into the one or more groups of customers, compute a lift of a signature of each of the one or more discriminatory sequence patterns as a ratio that the signature of a first one of the one or more discriminatory sequence patterns occurring in a first one of the classes also occurs in a second one of the one or more discriminatory sequence patterns in a second one of the classes, wherein the lift of the signature is used to determine when to compress the signature and when to discard the signature, and wherein timing information of the signature with respect to an event of interest is augmented to the signature such that the timing information is used to assign an importance to the signature as compared with an alternative signature notwithstanding whether the signature has a lower computed lift with respect to the alternative signature;
predict a response outcome of the customer journey for a user according to an assigned score based on the one or more discriminatory sequence patterns identified between the one or more of the groups of customers using the trained neural network; and
display the predicted response outcome of the customer journey on a display associated with the one or more computers.

8. The system of claim 7 , wherein the executable instructions further identify one or more friction points on of the customer journey.

9. The system of claim 7 , wherein the executable instructions further:
learn a successful response outcome of each customer journey generated from a first discriminatory sequence pattern associated with a first group of customers; or
learn an unsuccessful response outcome of each customer journey generated from a second discriminatory sequence pattern associ

Assignees

• IBM

Inventors

• Ganti Raghu Kiran
• Agrawal Dakshi
• Srivatsa Mudhakar
• Dewan Pranita

Classifications

• G06Q30/0202

  Market predictions or forecasting for commercial activities · CPC title

• G06Q30/0204 Primary

  Market segmentation · CPC title

• G06N20/00

  Machine learning · CPC title