Tracking of near conversions in user engagements

1. A computer implemented method for user analysis, comprising:

connecting a user and an agent to an interaction engine;

tracking interactions between the user and the agent with the interaction engine;

an analysis engine analyzing interactions between said user and said agent to determine why the user did not convert, and whether the user should be categorized as a near convert for potential retargeting, wherein near converted users are those non-converted users who have a high probability of being converted;

said analysis engine further comprising a text mining module that is configured for performing topic modeling to represent user and agent interaction transcripts in terms of a set of N topics;

wherein a topic is a distribution over a vocabulary that comprises all words in the transcripts;

wherein said text mining module analyzes said transcripts of interactions over a period of time to obtain a topic model comprising said N topics;

wherein on receiving a transcript, said text mining module analyzes said transcript and identifies the topics that are present in the transcript;

wherein said analysis engine further comprises a control module that builds a model in which sales conversion is a response variable, and in which topics received from said text mining module are independent variables;

wherein based on the identified topics present in the transcript, the control module scores the transcript;

said analysis engine, based on said score, categorizing the interaction between the user and the agent into categories;

flagging, based on the categories, the user for retargeting; and

said analysis engine, retargeting the user based on the flagging of the user, wherein said retargeting attempts to convert a user who is potentially interested in a good or service into a user who actually purchases the good or service by modifying the user's experience.

2. The method of claim 1 , wherein said interactions comprise inquiries about a product or service, and a service call.

3. The method of claim 1 , further comprising:

analyzing interactions between said user and said agent to determine, based on a threshold, whether the user is a near-convert.

4. The method of claim 1 , further comprising:

analyzing said interactions based on transcripts between said user and said agent.

5. The method of claim 1 , further comprising:

selecting categories for retargeting users who are near-converts via any of a plurality of channels.

6. A method for tracking near conversions in user engagements, comprising:

providing an interaction engine by which a user and an agent interact with each other, said interaction engine using any of a plurality of available channels as a mode of interaction;

providing an analysis engine for categorizing said user into a converted user or a non-converted user;

wherein a converted user is a user with whom the agent has been able to complete a transaction;

wherein a non-converted user is a user with whom the agent was unable to complete the transaction;

wherein said analysis engine is configured for accessing transcripts of interactions between said agent and said non-converted user in a textual format;

wherein said analysis engine is configured for analyzing said transcripts to identify near converted users, wherein near converted users are those non-converted users who have a high probability of being converted;

wherein said analysis engine is configured for analyzing journeys of a sample set of users, wherein said sample set of users comprises converted users, non-converted users, and users who have been designated as near converted users;

based on said analysis, said analysis engine builds a model that creates a correlation between the journeys and the type of user;

wherein when said interaction engine detects user interaction, said analysis engine maps a type of user to one of a potential converted user, a potential non-converted user, or a potential near converted user by mapping a current journey of the user to said model; and

said interaction engine retargeting near converted users, wherein said retargeting attempts to convert a user who is potentially interested in a good or service into a user who actually purchases the good or service by modifying the user's experience;

said analysis engine further comprising a text mining module that is configured for performing topic modeling to represent user and agent interaction transcripts in terms of a set of N topics;

wherein a topic is a distribution over a vocabulary that comprises all words in the transcripts;

wherein said text mining module analyzes said transcripts of interactions over a period of time to obtain a topic model comprising said N topics;

wherein on receiving a transcript, said text mining module analyzes said transcript and identifies the topics that are present in the transcript;

wherein said analysis engine further comprises a control module that builds a model in which sales conversion is a response variable, and in which topics received from said text mining module are independent variables;

wherein based on the identified topics present in the transcript, the control module scores the transcript;

wherein based on the score, the control module classifies the transcripts as near convert users;

wherein transcripts having a high score are those related to a near converted user; and

whereas transcripts having a low score are classified as these related to users with a low chance of being converted.

7. The method of claim 6 , further comprising:

providing a mechanism for transcribing a voice based interaction between said agent and said non-converted user into textual format.

8. The method of claim 6 , wherein said control module determines weights for each of the topics.

9. The method of claim 8 ,

wherein said text mining module identifies purchase and/or sale information that corresponds to the transcript by examining the disposition of the agent associated with the transcript, wherein the disposition of the agent comprises sale information in the transcript that indicates whether or not a sale was made.

10. A method for identifying a user as a near-converted user, comprising:

an analysis engine receiving one or more transcripts of user and agent interactions from an interaction engine;

said analysis engine checking if a user corresponding to the transcript has been converted by checking information present in the disposition of an agent responsible for the user interaction;

wherein if the user has not been converted, the analysis engine analyzes the transcripts and identifies topics present in the transcript;

said analysis engine further comprising a text mining module that is configured for performing topic modeling to represent user and agent interaction transcripts in terms of a set of N topics;

wherein a topic is a distribution over a vocabulary that comprises all words in the transcripts;

wherein said text mining module analyzes said transcripts of interactions over a period of time to obtain a topic model comprising said N topics;

wherein on receiving a transcript, said text mining module analyzes said transcript and identifies the topics that are present in the transcript;

wherein said analysis engine further comprises a control module that builds a model in which sales conversion is a response variable, and in which topics received from said text mining module are independent variables;

wherein based on the topics present in the transcript, the analysis engine scores the transcript;

wherein the analysis engine checks if the score is above a predetermined threshold;

wherein if the score is above a threshold, the analysis engine classifies the user corresponding to the transcript as a near converted user, wherein near converted users are those non-converted users who have a high probability of being converted; and

said interaction engine retargeting near converted users, wherein said retargeting attempts to convert a user who is potentially interested in a good or service into a user who actually purchases the good or service by modifying the user's experience.

11. The method of claim 10 , wherein users who have had no interaction with an agent, but who have undertaken a journey related to a product or a service by navigating across a Web site, are classified as near converted users, based on their journey.