US 12,277,568 B2
Systems and methods for determining lifetime value of website visitor through machine learning
Praveen Palaniyappan Vaikunta Raman, Los Angeles, CA (US); Sharat Nair, Torrance, CA (US); and Nicholas Sarnoff, Los Angeles, CA (US)
Assigned to TrueCar, Inc., Santa Monica, CA (US)
Filed by TRUECAR, INC., Santa Monica, CA (US)
Filed on Dec. 30, 2021, as Appl. No. 17/566,471.
Claims priority of provisional application 63/132,734, filed on Dec. 31, 2020.
Prior Publication US 2022/0207541 A1, Jun. 30, 2022
Int. Cl. G06Q 30/00 (2023.01); G06N 20/00 (2019.01); G06Q 30/0201 (2023.01)
CPC G06Q 30/0201 (2013.01) [G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method for improving search results using machine learning, the method comprising:
training a machine learning model, wherein the training of the machine learning model comprises:
receiving input data from at least one of: a data store or a cloud storage;
preparing datasets from the input data for training and testing the machine learning model, the datasets containing user-related features associated with lead submissions, dealer-related features associated with dealers, and vehicle-related features associated with vehicles;
dividing the datasets into a training dataset and a test dataset;
training the machine learning model using the training dataset, wherein the training produces first predictions of user values;
testing the machine learning model using the test dataset, wherein the testing produces second predictions of user values; and
comparing the first predictions and the second predictions; and
performing the training, the testing, and the comparing until a result of the comparing meets a performance criterion;
receiving, by a vehicle data system operating on a server machine, a lead submission from a user through a website supported by the vehicle data system, wherein the user becomes a lead through the lead submission and wherein the vehicle data system comprises the machine learning model;
determining, by the vehicle data system utilizing the trained machine learning model, a user value for the lead associated with the lead submission, wherein the user value represents a probability of the lead purchasing the vehicle from the dealer through the website supported by the vehicle data system;
determining, by the vehicle data system utilizing the trained machine learning model in conjunction with one or more factors, a user lifetime value for the lead based at least on the user value for the lead, the one or more factors comprising a multiplier;
obtaining, by the vehicle data system from a search engine on the Internet, clickstream identifiers, wherein each of the clickstream identifiers is associated with a respective visitor of a plurality of respective visitors of the website supported by the vehicle data system;
for each of the clickstream identifiers, assigning, by the vehicle data system utilizing the trained machine learning model, a corresponding user lifetime value for the respective visitor, wherein the machine learning model is trained to, for each lead submission, take as input user-related features associated with the respective visitor, dealer-related features associated with a dealer of interest, and vehicle-related features associated with a vehicle of interest and output a prediction that indicates how likely the respective visitor is to make a purchase of the vehicle of interest from the dealer of interest through the website supported by the vehicle data system;
aggregating, by the vehicle data system in a file, the clickstream identifiers and corresponding user lifetime values;
communicating, by the vehicle data system over the Internet, the file to a search server on which the search engine operates; and
conducting, by the search engine, improved searches utilizing the user lifetime values such that the search engine is able to provide improved search results in response to search requests from user devices.