Systems and methods for assessing, correlating, and utilizing online browsing and sales data

Patent metadata

Field	Value
Publication number	US-11727427-B2
Application number	US-202017080909-A
Country	US
Kind code	B2
Filing date	Oct 27, 2020
Priority date	Oct 27, 2020
Publication date	Aug 15, 2023
Grant date	Aug 15, 2023

Abstract

The present disclosure provides systems, methods, and metrics that filter out online visitor behavioral data that represents a potential lead with a high likelihood to convert to a vehicle sale from online visitor behavioral data that does not represent a potential lead with a low likelihood to convert to a vehicle sale, based on a mapping of sales back to observed website and vehicle configurator data. This enables more effective lead generation and the more efficient targeting of online incentive offers and sales “nudges,” for example. Further, the present disclosure enables web analytics data to be combined with sales data for sales forecasting in general.

First claim

What is claimed is:

1. A method, comprising:
receiving web analytics data comprising identification information from a web analytics database of a web interface;
filtering the web analytics data comprising the identification information to generate subset of the web analytics data that indicates a probability for completion of a sales transaction above a predetermined probability taking into account product configurations that do not correlate with any sales data;
offering users or sessions associated with the subset of the web analytics data an offer inducement to complete sales transaction via the web interface;
correlating historical sales transaction data to the web analytics data to determine which product types are more probable to be involved in the sales transaction and which product types are less probable to be involved in the sales transaction; and
adjusting the product configurations of one or more of a manufacturing operation and an inventory allotment based on the determination as to which product types are more probable to be involved in the sales transaction and which product types are less probable to be involved in the sales transaction;
wherein correlating the historical sales transaction data to the web analytics data comprises:
encoding the web analytics data and the historical sales transaction data such that a resulting feature space represents product configurations, price per product configuration, and a fraction of each product configuration sold in a time period, T, for machine language processing by a neural network;
clustering web analytics records associated with the web analytics data and sales transaction records associated with the sales transaction data;
for each cluster, assigning a nearest cluster identification using a k-nearest neighbor method;
establishing a cost function between clusters;
selecting a training cluster and a plurality of validation clusters;
training the neural network using the training cluster and validating output of the neural network using the plurality of validation clusters;
based on the output of the neural network, generating and displaying on a display a joint distribution heat map indicating correlations between the web analytics data and the historical sales transaction data; and
using the joint distribution heat map, identifying the product configurations that do not correlate with any sales data.

2. The method of claim 1 , wherein filtering the web analytics data comprising the identification information comprises appending usage-specific feature data to each record of the web analytics data comprising the identification information.

3. The method of claim 2 , wherein the usage-specific feature data comprises one or more of: (1) number of unique sets of product configurations established by each user using the web interface, (2) number of product types explored by each user using the web interface, (3) price range explored by each user using the web interface, (4) number of unique sessions engaged in by each user to arrive at a given product configuration using the web interface, and (5) total time spent by each user to arrive at the given product configuration.

4. The method of claim 2 , wherein filtering the web analytics data comprising the identification information further comprises determining a probability for the completion of the sales transaction for each record of the web analytics data comprising the identification information and the appended usage-specific feature data using a classification data model and thresholding the probability for the completion of the sales transaction for each record of the web analytics data comprising the identification information.

5. The method of claim 4 , wherein the classification data model comprises one of a neural network model, a random forest decision tree model, and a k-nearest neighbor model.

6. A non-transitory computer-readable medium comprising instructions stored in a memory and executed by a processor to carry out the steps comprising:
receiving web analytics data comprising identification information from a web analytics database of a web interface;
filtering the web analytics data comprising the identification information to generate subset of the web analytics data that indicates a probability for completion of a sales transaction above a predetermined probability taking into account product configurations that do not correlate with any sales data;
offering users or sessions associated with the subset of the web analytics data an offer inducement to complete sales transaction via the web interface;
correlating historical sales transaction data to the web analytics data to determine which product types are more probable to be involved in the sales transaction and which product types are less probable to be involved in the sales transaction; and
adjusting the product configurations of one or more of a manufacturing operation and an inventory allotment based on the determination as to which product types are more probable to be involved in the sales transaction and which product types are less probable to be involved in the sales transaction;
wherein correlating the historical sales transaction data to the web analytics data comprises:
encoding the web analytics data and the historical sales transaction data such that a resulting feature space represents product configurations, price per product configuration, and a fraction of each product configuration sold in a time period, T, for machine language processing by a neural network;
clustering web analytics records associated with the web analytics data and sales transaction records associated with the sales transaction data;
for each cluster, assigning a nearest cluster identification using a k-nearest neighbor method;
establishing a cost function between clusters;
selecting a training cluster and a plurality of validation clusters;
training the neural network using the training cluster and validating output of the neural network using the plurality of validation clusters;
based on the output of the neural network, generating and displaying on a display a joint distribution heat map indicating correlations between the web analytics data and the historical sales transaction data; and
using the joint distribution heat map, identifying the product configurations that do not correlate with any sales data.

7. The non-transitory computer readable medium of claim 6 , wherein filtering the web analytics data comprising the identification information comprises appending usage-specific feature data to each record of the web analytics data comprising the identification information.

8. The non-transitory computer readable medium of claim 7 , wherein the usage-specific feature data comprises one or more of: (1) number of unique sets of product configurations established by each user using the web interface, (2) number of product types explored by each user using the web interface, (3) price range explored by each user using the web interface, (4) number of unique sessions engaged in by each user to arrive at a given product configuration using the web interface, and (5) total time spent by each user to arrive at the given product configuration.

9. The non-transitory computer readable medium of claim 7 , wherein filtering the web analytics data comprising the identification information further comprises determining a probability for the completion of the sales transaction for each record of the web analytics data comprising the identification information and the appended usage-specific feature data using a classification data model and thresholding the probability for the completion of the sales transaction for each record of the web analytics data

Assignees

• Volvo Car Corp

Inventors

• Roy Chowdhury Sohini
• Seneger Jon
• Alareqi Ebrahim
• Soderberg Joakim
• Liu Ao

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06Q30/0222 Primary

  During e-commerce, i.e. online transactions · CPC title

• G06F16/9536

  Search customisation based on social or collaborative filtering · CPC title

• G06N3/08

  Learning methods · CPC title