# US 2017/0255945 A1 · System and method for intelligent sales engagement

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2017/0255945 A1
- Document ID: US20170255945A1
- Status: Published application
- Kind: A1
- Application number: 15/393,978
- Filed: 2016-12-29
- Published: 2017-09-07
- Earliest claimed: 2016-03-07
- Assignees: NewVoiceMedia, Ltd.
- Applicants: NewVoiceMedia Ltd.
- Inventors: Alan McCord; Ashley Unitt; Mark Fellowes; Andrew Carson; Selma Ardelean
- Web page: https://patentgrep.com/patent/US20170255945A1

## CPC

- G06Q 30/0201: Market modelling; Market analysis; Collecting market data
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06Q 10/067: Enterprise or organisation modelling
- G06Q 10/1097: {using calendar-based scheduling for task assignment}
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A system for automatically automatic workflow triggering using real-time analytics, comprising an analytics server that receives and analyzes interaction information and a workflow server that produces workflow events based on the analysis, sends workflow events to handlers for processing, retrieves workflow-related data, and produces workflow reports for review, and a method for automatically automatic workflow triggering using real-time analytics.

## Claims

### Claim 1

A system for intelligent sales engagement, comprising:
a pre-integrated graph module comprising at least a plurality of programming instructions stored in a memory and operating on a processor of a network-connected computing device;
a machine learning module comprising at least a plurality of programming instructions stored in a memory and operating on a processor of a network-connected computing device; and
an optimization module comprising at least a plurality of programming instructions stored in a memory and operating on a processor of a network-connected computing device;
wherein, the pre-integrated graph module:
(a) monitors and captures events from source systems and constructs an event graph of multichannel interactions and attributes including firm demographics and sales rep attributes;
(b) automatically reduces the graph to the significant state transitions, effectively reverse engineering the sales process from available actual event data;
(c) runs in modes where reducing the graph happens either periodically or continuously; and
(d) supports different pre-defined topologies of funnel, decision circle and journey;
wherein the machine learning module:
(e) trains a family of predictive machine-learning models (ex. Distributed Random Forest, Deep Learning) for any transition of interest (or for all transitions) of the reduced graph and performs validation of the accuracy (AUC) of each predictive machine learning model;
(f) chooses different model types for different transitions based on model with highest accuracy, model then estimates the conditional probability of the transition from the starting to ending state potentially including all known input attributes at the starting state;
(g) accepts case determined numbers of input attributes of different types on each state transitions;
(h) runs with either full state history with attributes, Markov approximation or hidden Markov model or a hybrid mode; and
(i) supports a hidden Markov model to represent the hidden “intent” state of the contact or lead;
wherein the optimization module:
(j) creates a set of visualizations showing the various resulting performance metrics including conversion rate, representative utilization percentage, and total value in the pipeline;
(k) uses the trained predictive models as input to an automated optimization phase which recommends specific actions (interactions) to take to optimize the business outcome of prospects flowing through the reduced graph subject to constraints;
(l) supports optimization under uncertainty;
(m) schedules interactions between agents and prospects to maximize an objective; and
(n) configures, in addition to existing model optimization, optimization experiments that are executed and is then able to run experiments, analyze the results and self-learn giving rise to increased utility.

### Claim 2 (depends on 1)

The system of claim 1, whereas the expected sales process may be entered as input to guide graph reduction and/or highlight deviations from expected flows.

### Claim 3 (depends on 1)

The system of claim 1, wherein the graph may also represent B2B flows and B2C flows.

### Claim 4 (depends on 1)

The system of claim 1 wherein the machine learning module may learn or reverse engineer a process based on historical data.

### Claim 5 (depends on 1)

The system of claim 1 wherein, the machine learning module may account for the multi-dimensional nature of social influence, and the role of advocates who aren't customers.

### Claim 6 (depends on 1)

The system of claim 1 wherein, the machine learning module may shift to ongoing relationships beyond individual transactions.

### Claim 7 (depends on 1)

The system of claim 1 wherein, the machine learning module runs in an adaptive mode where retraining happens periodically or continuously.

### Claim 8

A method for intelligent sales engagement, the method comprising the steps of:
(a) monitoring and extracting sets of customer relationship sales data from source systems into a pre-integrated graph module comprising at least a plurality of programming instructions stored in a memory and operating on a processor of a network-connected computing device;
(b) constructing an event driven relational graph of multichannel interactions and attributes including firm demographics and sales rep attributes using the pre-integrated graph module;
(c) reducing the graph to the significant state transition occurrences, effectively reverse engineering the sales process from available actual event data expressing the resultant graph in one of a plurality of pre-defined topologies such as: funnel, decision circle and engagement journey using the pre-integrated graph module;
(d) training a family of predictive machine-learning models (e.g. Distributed Random Forest, deep learning, etc.) for any transition of interest (or for all transitions) of the reduced graph and performs validation of the accuracy (AUC) of each predictive machine learning model using a machine learning module comprising at least a plurality of programming instructions stored in a memory and operating on a processor of a network-connected computing device;
(e) choosing different model types using the machine learning module for different transitions based on model with highest accuracy, chosen model then estimates conditional probability of the transition from the starting to ending state potentially including all known input attributes at the starting state;
(f) running either full state history with attributes, Markov approximation or hidden Markov model or a hybrid mode using the machine learning module;
(g) creating a set of visualizations showing the various resulting performance metrics including conversion rate, rep utilization %, and total value in the pipeline using the optimization module;
(h) using the trained predictive models as input to an automated optimization phase which recommends specific actions (interactions) to take to optimize the business outcome of prospects flowing through the reduced graph subject to constraints using the optimization module.

### Claim 9 (depends on 8)

The method of claim 8, wherein the expected sales process may be entered as input to guide graph reduction and to highlight deviations from expected flows.

### Claim 10 (depends on 8)

The method of claim 8, wherein the graph may also represent B2B flows and B2C flows.

### Claim 11 (depends on 8)

The method of claim 8, wherein the machine learning module may learn or reverse engineer a process based on historical data.

### Claim 12 (depends on 8)

The method of claim 8, wherein the machine learning module may account for the multi-dimensional nature of social influence, and the role of advocates who aren't customers;

### Claim 13 (depends on 8)

The method of claim 8, wherein the machine learning module may shift to ongoing relationships beyond individual transactions.

### Claim 14 (depends on 8)

The method of claim 8, wherein the machine learning module runs in an adaptive mode where retraining happens periodically or continuously.
