Granted Patent B2

US 12,456,132 · App. 18/620,239 · Granted Oct 28, 2025

Inventors: Pavan Korada (San Rafael, CA); Sunpreet Singh Khanuja (San Jose, CA); Yun Sam Chong (Santa Clara, CA); Bharat Goyal (San Jose, CA); Edward Robert Rau, Jr. (Riverbank, CA)

Assignee: Zeta Global Corp.

G06Q30/0243G06F16/24578G06F16/9535G06F30/20G06Q30/016G06Q30/0251

Abstract

Claims (60)

1 . An adaptive real time modeling and scoring system for generating predictive scoring models, the system comprising:

one or more processors; and

a memory coupled to the one or more processors which stores processor-executable instructions which, when executed by the one or more processors, cause the one or more processors to:

generate multiple predictive scoring models;

identify a consolidated trigger for a combination of the multiple predictive scoring models;

evaluate the multiple predictive scoring models based on a heuristic to select an optimal predictive scoring model;

determine, based on the consolidated trigger, that the optimal predictive scoring model is ready for a refresh;

identify a change in system input data, the system input data comprising elements of a plurality of consumer profiles including one or more of a plurality of email traffic, display traffic and social media traffic;

refresh the predictive scoring model based on the identified change in system input data;

use the refreshed predictive scoring model for generating a look-alike audience; and

generate the look-alike audience to improve system efficiency and data accuracy in audience targeting at least by automatically discovering new data relationships and trends as historical data evolves over time.

2 . The system of claim 1 , wherein the refreshed predictive scoring model is for calculating new lead scores indicating a probability that the new leads will make a purchase.

3 . The system of claim 1 , wherein the processor-executable instructions, when executed, further cause the one or more processors to receive an input defining a consumer profile of an existing target consumer; and

generate a target look-alike audience comprising potential target consumers replicating, at least in part, aspects of the consumer profile.

4 . The system of claim 3 , wherein the processor-executable instructions, when executed, further cause the one or more processors to receive user selections relating to at least some aspects of the consumer profile with an interactive user interface.

5 . The system of claim 4 , wherein the processor-executable instructions, when executed, further cause the one or more processors to receive a selection of a degree of replication accuracy or population size of the target look-alike audience with a consumer element of the interactive user interface.

6 . The system of claim 1 , wherein the processor-executable instructions, when executed, further cause the one or more processors to:

determine an equation including multiple predictive factors weighted by an importance of each predictive factor in predicting a likelihood of a lead to convert; and

change at least one weight of a predictive factor to adapt the refreshed predictive scoring model to include a quality classification for a publisher.

7 . The system of claim 1 , wherein the consolidated trigger is an automated trigger that is based on analysis of historical data and new data that reveals a new relationship between multiple variables that did not exist previously.

8 . A method for performing adaptive real time modeling and scoring, the method comprising, at least:

generating multiple predictive scoring models;

identifying a consolidated trigger for a combination of the multiple predictive scoring models;

evaluating the multiple predictive scoring models based on a heuristic to select an optimal predictive scoring model;

determining, based on the consolidated trigger, that the optimal predictive scoring model is ready for a refresh;

identifying a change in system input data, the system input data comprising elements of a plurality of consumer profiles including one or more of a plurality of email traffic, display traffic and social media traffic;

refreshing predictive scoring model based on the identified change in system input data;

using the refreshed predictive scoring model for one or more of recommending strategies for display media bidding; and

recommending the one or more strategies for display media bidding to improve system efficiency and data accuracy in audience targeting at least by automatically determining differences between pricing assumptions and at least one of an observed quality metric and bid performance.

9 . The method of claim 8 , wherein the refreshed predictive scoring model is for calculating new lead scores indicating a probability that the new leads will make a purchase.

10 . The method of claim 8 , further comprising:

receiving an input defining a consumer profile of an existing target consumer; and

generate a target look-alike audience comprising potential target consumers replicating, at least in part, aspects of the consumer profile.

11 . The method of claim 10 , further comprising:

providing a look-alike audience creator, the look-alike audience creator including an interactive user interface for receiving user selections relating to at least some aspects of the consumer profile.

12 . The method of claim 11 , further comprising:

using the interactive user interface to receive a selection of a degree of replication accuracy or population size of the target look-alike audience.

13 . The method of claim 8 , further comprising:

determining an equation including multiple predictive factors weighted by an importance of each predictive factor in predicting a likelihood of a lead to convert; and

changing at least one weight of a predictive factor to adapt the refreshed predictive scoring model to include a quality classification for a publisher.

14 . The method of claim 8 , wherein the consolidated trigger is an automated trigger that is based on analysis of historical data and new data that reveals a new relationship between different variables that did not exist previously.

15 . A method for performing adaptive real time modeling and scoring, the method comprising, at least:

generating multiple predictive scoring models;

identifying a forced trigger for a combination of the multiple predictive scoring models;

evaluating the multiple predictive scoring models based on a heuristic to select an optimal predictive scoring model;

determining, based on the forced trigger, that the optimal predictive scoring model is ready for a refresh;

identifying a new data file at a pre-defined server location, the new data file comprising an advertiser dataset including elements of a plurality of consumer profiles, each of the plurality of consumer profiles including one or more of a plurality of email traffic, display traffic and social media traffic;

regenerating the predictive scoring model based on the identified new data file;

using the regenerated predictive scoring model for one or more of recommending strategies for display media bidding; and

recommending the one or more strategies for display media bidding to improve system efficiency and data accuracy in audience targeting at least by automatically determining differences between pricing assumptions and at least one of an observed quality metric and bid performance.

16 . The method of claim 15 , wherein the regenerated predictive scoring model is for calculating new lead scores indicating a probability that the new leads will make a purchase.

17 . The method of claim 15 , further comprising:

receiving an input defining a consumer profile of an existing target consumer, and generating a target look-alike audience comprising potential target consumers replicating, at least in part, aspects of the consumer profile.

18 . The method of claim 17 , further comprising:

providing a look-alike audience creator, the look-alike audience creator including an interactive user interface for receiving user selections relating to at least some aspects of the consumer profile.

19 . The method of claim 18 , further comprising:

using the interactive user interface to receive a selection of a degree of replication accuracy or population size of the target look-alike audience.

20 . The method of claim 15 , further comprising:

determining an equation including multiple predictive factors weighted by an importance of each predictive factor in predicting a likelihood of a lead to convert; and

changing at least one weight of a predictive factor to adapt the refreshed predictive scoring model to include a quality classification for a publisher.

Assignments (1)

Continuity (4)

References Cited (109)