Intelligent prediction systems and methods for conversational outcome modeling frameworks for sales predictions

Patent metadata

Field	Value
Publication number	US-11829920-B2
Application number	US-202016927321-A
Country	US
Kind code	B2
Filing date	Jul 13, 2020
Priority date	Jul 13, 2020
Publication date	Nov 28, 2023
Grant date	Nov 28, 2023

Abstract

An intelligent prediction system includes one or more processors, one or more memory components, and machine-readable instructions that cause the intelligent prediction system to: receive text data comprising a plurality of speaker turn segments of a transcription of a conversation, each speaker turn segment of the plurality of speaker turn segments representative of a turn in the conversation, the plurality of speaker turn segments collectively representative of the conversation up to a point of time, generate a point in time bind probability based on a speaker turn segment bind probability of a speaker turn segment at the point in time and memory data associated with the plurality of segments up to the point in time, and generate a speaker turn segment impact score at the point in time by subtracting an immediately preceding point in time bind probability from the point in time bind probability.

First claim

What is claimed is:

1. An intelligent prediction system for conversational outcome prediction, comprising:
one or more processors;
one or more memory components communicatively coupled to the one or more processors; and
machine-readable instructions stored in the one or more memory components that cause the intelligent prediction system to perform at least the following when executed by the one or more processors:
receive text data comprising a plurality of speaker turn segments of a transcription of a conversation between two or more individuals regarding a sale offer, each speaker turn segment of the plurality of speaker turn segments of the transcription representative of a turn in the conversation associated with speech data of one of the two or more individuals, the plurality of speaker turn segments collectively representative of the conversation up to a point of time;
vectorize each speaker turn segment of the text data to assign an associated numerical value to each speaker turn segment;
apply a padding function to each speaker turn segment with the vectorization such that each speaker turn segment comprises an equivalent segment length, wherein the padding function comprises addition of zeros to adjust each speaker turn segment to the equivalent segment length;
apply a loss function configured to minimize error to each speaker turn segment after vectorization;
generate, via a neural network prediction model, a point in time bind probability representative of a likelihood of a successful outcome of the sale offer at the point in time based on (i) a speaker turn segment bind probability of a speaker turn segment at the point in time and (ii) memory data associated with the plurality of speaker turn segments up to the point in time;
generate, via the neural network prediction model, a speaker turn segment impact score at the point in time by subtracting an immediately preceding point in time bind probability from the point in time bind probability; and
train the neural network prediction model based on a plurality of data sets of pre-stored conversations, wherein the neural network prediction model is updated and refined as new conversations become available to identify patterns indicative of increasing and decreasing bind probabilities.

2. The intelligent prediction system of claim 1 , further comprising:
a display communicatively coupled to the one or more processors;
wherein the machine-readable instructions further cause the intelligent prediction system to
generate a point in time bind probability plot to graphically display the point in time bind probability at each turn of the conversation, and
display the point in time bind probability plot on the display.

3. The intelligent prediction system of claim 1 , wherein the machine-readable instructions further cause the intelligent prediction system to:
identify one or more flag events corresponding to a predetermined sale technique; and
analyze each speaker turn segment to identify the one or more flag events corresponding to the predetermined sale technique.

4. The intelligent prediction system of claim 3 , further comprising:
a display communicatively coupled to the one or more processors, wherein the machine-readable instructions further cause the intelligent prediction system to:
generate a point in time bind probability plot to graphically display the point in time bind probability at each turn of the conversation on the display;
generate one or more markers associated with the one or more flag events that are identified at each turn in which the one or more flag events occurred; and
provide a visual output on the display of the one or more markers on the point in time bind probability plot.

5. The intelligent prediction system of claim 3 , wherein the one or more flag events comprise at least one of:
a customer service representative asking for a sale;
a customer asking for a sale; and
the customer service representative offering a discount.

6. The intelligent prediction system of claim 1 , wherein the machine-readable instructions further cause the intelligent prediction system to:
receive customer data from one or more customer information sources;
determine a starting bind probability based on the customer data; and
generate the point in time bind probability representative of the likelihood of the successful outcome of the sale offer at the point in time based on the starting bind probability.

7. The intelligent prediction system of claim 1 , wherein the machine-readable instructions further cause the intelligent prediction system to:
determine an average bind probability based on one or more stored conversations; and
determine a set of performance metrics of a participant in the conversation, each performance metric of the participant determined based on a participant performance score in the conversation at each turn associated with the participant relative to the average bind probability;
generate a feedback performance metric for the participant based on the set of performance metrics;
compare in a comparison the average bind probability to the feedback performance metric of the participant in the conversation; and
generate a positive participant performance score when the comparison is positive such that the feedback performance metric is above the average bind probability.

8. The intelligent prediction system of claim 1 , wherein the machine-readable instructions further cause the intelligent prediction system to:
apply the loss function to each speaker turn segment after vectorization such that each speaker turn segment comprises a weight adjustment to achieve a weight adjustment value, wherein the loss function comprises adjusting each speaker turn segment based on the weight adjustment to achieve the weight adjustment value.

9. An intelligent prediction system for conversational outcome prediction, comprising:
one or more processors;
one or more memory components communicatively coupled to the one or more processors; and
machine-readable instructions stored in the one or more memory components that cause the intelligent prediction system to perform at least the following when executed by the one or more processors:
receive, via an audio capture module, audio data of a conversation;
transcribe the audio data of the conversation from the audio capture module into text data comprising a plurality of speaker turn segments of the conversation between two or more individuals regarding a sale offer, each speaker turn segment of the plurality of speaker turn segments of the transcription representative of a turn in the conversation associated with speech data of one of the two or more individuals, the plurality of speaker turn segments collectively representative of the conversation up to a point of time;
vectorize each speaker turn segment of the text data to assign an associated numerical value to each segment;
apply a padding function to each speaker turn segment with the vectorization such that each speaker turn segment comprises an equivalent segment length, wherein the padding function comprises addition of zeros to adjust each speaker turn segment to the equivalent segment length;
apply a loss function configured to minimize error to each speaker turn segment post after vectorization;
generate, via a neural network prediction model, a point in time bind probability representative of a likelihood of a successful outcome of the sale offer at the point in time based on (i) a speaker turn segment bind probability of a speaker turn segment at the point in time and (ii) memory data associated with the plurality of speaker turn segments up to the point in time;
generate, via the neural network prediction mode

Assignees

• Allstate Insurance Co

Inventors

• Pripstein Eric
• Fiddler Garrett

Classifications

• G06Q10/06398 Primary

  Performance of employee with respect to a job function · CPC title

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G06Q30/0202

  Market predictions or forecasting for commercial activities · CPC title

• G10L15/063

  Training · CPC title

• G10L15/16

  using artificial neural networks · CPC title