METHOD AND APPARATUS FOR INTENT MODELING AND PREDICTION

• US 20130268468A1
• Filed: 03/28/2013
• Published: 10/10/2013
• Est. Priority Date: 04/09/2012
• Status: Active Grant

• Current Assignee: [24]7.ai, Inc.
• Original Assignee: 24/7 Customer Incorporated
• Inventors: VIJAYARAGHAVAN, Ravi, KULKARNI, Subhash Ramchandra, ADUSUMILLI, Kranthi Mitra
• Granted Patent: US 9,275,342 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/06 Buying, selling or leasing transactions
• CPC Class Codes: G06Q 30/0613 Third-party assisted
• CPC Class Codes: G06Q 30/0631 Item recommendations

Abstract

A method and apparatus enables identification of customer characteristics and behavior, and predicts the customer'"'"'s intent. Such prediction can be used to adopt various business strategies to increase the chances of conversion of customer interaction to a sale, and thereby can increase revenue, and/or enhance the customer'"'"'s experience.

Claims

1. A computer implemented method for addressing needs of a visitor accessing a website, comprising:

a processor collecting information about each visitor accessing the website and storing the information in a database;

said processor training a visitor intent model based on historical visitor information stored in the database;

said processor collecting information about a particular visitor;

said processor determining a group with which to associate the particular visitor based on the collected information about the particular visitor; and

after the visitor intent model is trained, said processor predicting intent of a particular visitor accessing the website based on information about the particular visitor and the visitor intent model.

2. The method of claim 1, further comprising:

said processor determining an engagement strategy for helping a particular visitor to be successful in achieving the predicted intent; and

said processor engaging the particular visitor using the determined engagement strategy.

3. The method of claim 1, wherein the information collected about each visitor comprises any of:

location of the visitor;

day and time at which the visitor accessed the website;

length of time the visitor interacts with the website;

past purchase history;

source of referral to the website comprising any of email, search engines, social media, and other websites;

device used for interaction;

operating system and version;

Internet browser and version;

IP address;

first page within the website accessed by the visitor;

log of Web accesses within the website;

online chat transcript;

voice call transcripts; and

visitor interaction information.

4. The method of claim 1, wherein predicting intent of the particular visitor comprises:

said processor executing the visitor intent model to determine the predicted intent of the particular visitor using the group associated with the particular visitor and the information collected about the particular visitor.

5. The method of claim 2, wherein the engagement strategy comprises any of:

offering proactive recommendations;

offering chat services;

initiating a telephone call to the visitor;

inviting a telephone call from the visitor;

providing a customer chat and/or widget;

offering a discount; and

providing a promotional offer.

6. The method of claim 1, wherein the intent of the particular visitor comprises any of:

browse the website;

collect information for investigation or research;

purchase a product;

check status on a purchase order;

make a payment;

dispute a payment;

issue resolution;

schedule an appointment; and

schedule an on-site service call.

7. A computer implemented method for assessing visitor interactions with a website, comprising:

a processor implementing an intent prediction engine for receiving information from a visitor accessing said website;

said intent prediction engine segmenting visitors to said website into one or more of a plurality of groups based on a plurality of factors;

said intent prediction engine developing models for visitor behavior for each of said groups;

on detecting a visitor accessing said website, said intent prediction engine identifying characteristics of said visitor, categorizing said visitor into one or more of said groups, modeling said visitor'"'"'s behavior with respect to a previous population of visitors in the relevant groups identified, and predicting an intention of the visitor based on modeling of said visitor'"'"'s behavior.

8. The method of claim 7, wherein based upon said predicted intent, said intent prediction engine proactively offering personalized recommendations and/or best multichannel support to said visitor.

9. The method of claim 7, wherein said intent prediction engine comprises:

a group identification module for identifying suitable homogenous groups of online visitors based on factors comprising any of location of the visitor, referral source, landing page and/or exit page, day and time of visit, and visitor interaction information, past purchase history; and

wherein said group identification module categorizes each visitor into one or more pre-defined groups of visitors by modeling the visitor behavior with respect to previous population of visitors in the relevant groups identified;

an intent prediction and business rules engine for using machine learning models to identify a visitor'"'"'s intent, once the visitor has been grouped into one of the identified groups;

a modeling engine for developing said models for visitor behavior for each of said groups and for storing said models in said prediction engine database;

wherein said modeling engine provides models to target appropriate visitors at an appropriate time by offering intervention strategies; and

said prediction engine database storing visitor information in a structured format, said visitor information comprising any of locations of visitors, time of visit, referral page, landing and exit pages, visitor interaction information, past purchase history, and chat and voice-call transcripts from e-commerce websites and other sources.

10. The method of claim 9, wherein an intent group comprises a set of visitors who have a same predicted intent, have exhibited a similar behavior in some aspect, and have a common attribute.

11. The method of claim 9, wherein the intent predicted by said rules engine comprises any of:

the visitor wants to perform a search to gather information about certain products;

the visitor wants to browse to gain knowledge;

the visitor wants to compare various products and accessories;

the visitor wants to purchase certain products; and

the visitor wants to update any of account details, shopping cart related information, and payment related information.

12. The method of claim 9, wherein based on a partial Web journey of the visitor, said modeling engine models a next action of the visitor as the visitor performs the Web journey, and at each stage, the modeling engine decides what appropriate pre-emptive action and/or engagement strategy can be taken to maximize metrics of interest.

13. The method of claim 9, wherein said modeling engine computes channel affinity for each couplet by computing a channel friction score and selects a suitable engagement strategy based on the computed channel friction score.

14. The method of claim 9, wherein said modeling engine selects an appropriate time to trigger an engagement strategy, based on a plurality of factors comprising, but not limited to, the time and/or time range that the visitor spends on a Web page in a website.

15. The method of claim 14, wherein said modeling engine selects more than one engagement strategy.

16. The method of claim 9, wherein said engagement strategy comprising any of:

offering personalized recommendations;

offering a better mode of multichannel support; and

offering discounts and promotional offers.

17. The method of claim 9, wherein said intent prediction engine comprises a feedback mechanism, wherein data related to a visitor, an engagement strategy chosen, and an outcome and/or response is analyzed and used for further refinement.

18. The method of claim 9, said intent modeling engine, based on a predicted intent, proactively suggesting suitable personalized engagement actions in the form of recommendations and/or best multichannel support to the visitor.

19. An apparatus for addressing needs of a visitor accessing a website, comprising:

a processor configured for collecting information about each visitor accessing the website and storing the information in a database;

said processor configured for training a visitor intent model based on historical visitor information stored in the database;

said processor configured for, after the visitor intent model is trained, predicting intent of a particular visitor accessing the website based on information about the particular visitor and the visitor intent model;

said processor configured for determining an engagement strategy for helping a particular visitor to be successful in achieving the predicted intent; and

said processor configured for engaging the particular visitor using the determined engagement strategy.

20. An apparatus for assessing visitor interactions with a website, comprising:

a processor configured for implementing an intent prediction engine for receiving information from a visitor accessing said website;

said intent prediction engine segmenting visitors to said website into one or more of a plurality of groups based on a plurality of factors comprising any of locations of the visitors, time of visit, referral page, landing and/or exit pages, and visitor interaction information;

said intent prediction engine developing models for visitor behavior for each of said groups;

on detecting a visitor accessing said website, said intent prediction engine identifying characteristics of said visitor, categorizing said visitor into one or more of said groups, modeling said visitor'"'"'s behavior with respect to a previous population of visitors in the relevant groups identified, and predicting an intention of the visitor based on modeling of said visitor'"'"'s behavior.

21. The apparatus of claim 20, said intent prediction engine comprising:

a group identification module for identifying suitable homogenous groups of online visitors based on factors comprising any of location of the visitor, referral source, landing page and/or exit page, day and time of visit, and visitor interaction information; and

wherein said group identification module categorizes each visitor into one or more pre-defined groups of visitors by modeling the visitor behavior with respect to previous population of visitors in the relevant groups identified;

an intent prediction and business rules engine for using dynamic models to identify a visitor'"'"'s intent, once the visitor has been grouped into one of the identified groups;

a modeling engine for developing said dynamic models for visitor behavior for each of said groups and for storing said models in said prediction engine database;

wherein said modeling engine provides models to target appropriate visitors at an appropriate time by offering intervention strategies; and

said prediction engine database storing visitor information in a structured format, said visitor information comprising any of locations of visitors, time of visit, referral page, landing and exit pages, visitor interaction information, data, weblog, and chat and voice-call transcripts from e-commerce websites and other sources.