Conversations History

The Conversation History module allows you to review the transcripts of past conversations and label certain conversations for follow-up or further action.

The Advanced Conversation Analytics feature provides a more detailed view of the conversation, including events and custom tags, which can help you better understand the interaction between the customer and the virtual assistant. Custom Tags can be reused across conversations for training purposes and for identifying areas where the virtual assistant could improve.

Note: The Conversations History Dashboard is available only post the 9.3 release, i.e. post-July 24, 2022.

Navigating to the Conversation History Dashboard

To access the Conversation History Dashboard, follow these steps:

1. Sign in to the Kore.ai Platform and select the required Virtual Assistant.
2. Click the Analyze tab on the top menu.
3. Click Conversation History on the left navigation menu as shown below.

Filter Views Page

The Filter Views page is the preface or landing page displayed when you click the Conversations History left menu item. You can create and view prebuilt and custom filters, filter configurations, and descriptions on this page in addition to the following:

• View the available conversation filters and their configurations.
• Create a custom filter to view only the required data.
• Access the Conversations History Dashboard for the selected preset.

Prebuilt Filters for Conversations History

As an analyst, you may need a customized view of the conversation’s history data based on specific conditional filters. This helps analyze conversations based on priority/criticality.

The Platform now supports Prebuilt Filters or Presets for the most common use case scenarios. Presets reduce the developer’s efforts in skimming through each conversation to understand and tag them for a specific use case.

How it works

1. The system checks the configuration parameters and the filter conditions for a conversation session.
2. The conversation session is mapped to the respective filter automatically.
3. Based on the date range selected on the Conversations History Dashboard, the system fetches and displays the conversations that meet the Prebuilt Filters’ criteria.

Types of Prebuilt Filters

The following table displays each preset’s filter name, predefined configuration parameters, and description. For more information on the configuration parameters, please click here.

Feedback Prebuilt Filter

The Platform supports the Feedback filter on the Filter Views panel to create custom feedback filters. Based on the filter criteria, the system groups and displays the conversations for the following parameters:

• Feedback Survey Template Type: Indicates if the selected type is NPS, CSAT, or Thumbs-up / Thumbs-down. Learn more.
• Operator: The conditional/logical operator that applies to the Feedback filter evaluation which includes one of the following:
  • Equals to
  • Less than or equals to
  • Greater than equals to
  • Greater than
  • Less than
• Value: The feedback survey score used for the conditional logic evaluation.

View Prebuilt Filters

To view the prebuilt filters, follow the steps below:

1. Navigate to Analyze → Conversations History on the left menu.
2. The prebuilt filters are categorized and listed on the Filter Views page.



3. Click Read More to see the configuration details of each conversation filter type.



4. Click any prebuilt filter name to access the Conversations History panel where relevant data is displayed.

Create a Custom Conversation Filter

A Conversation Filter helps group and display selective conversations history data based on a specific criterion on the dashboard. To add a custom filter, follow the steps below:

1. Click Analyze on the top menu.
2. Click Conversations History on the left menu.
3. On the Conversations History page, click + Create Filter.



4. On the Create Conversation Filter panel, provide inputs for the following fields:

4. Click Save.

The new custom filter is listed on the Filter Views page. The conversations are grouped and displayed Conversations History Dashboard based on the inputs provided on the Filter Views page.

Edit a Conversation Filter

You can edit one or more fields for an existing conversation filter. To edit a conversation filter, follow the steps below:

1. Click the custom filter on the Filter Views page.
2. Click the More Filters dropdown list on the Conversations History panel.



3. Modify the required fields on the Edit Conversation Filter window.
4. Click Apply.

Filter Criteria

You can customize the Conversations History data view by selecting the Filter Criteria on the dashboard. Please click here to view the available filters.

Conversations History Dashboard

The Conversations History Dashboard displays the following information for each conversation session that meets the defined filter condition(s) for the selected date range.

Containment Type

The following containment types affect the conversation flow and are displayed on top of individual Conversations History details panel:

• Self-service: Indicates that the conversation (initiated by the user) with a VA was successfully completed.
• Agent Transfer: Indicates that the conversation was transferred to a live agent using the Agent Transfer node in the Dialog Flow.
• User Drop-off: Indicates that the user stopped participating in the conversation before it was completed, either due to an error in the flow or some other reason.



Label

A custom label helps identify conversations that require follow-up or indicate an action item for an analyst during review.



Add a Custom Label

To add a custom label in addition to the default label for a conversation, follow the steps below:

1. Click the + Label button on the Conversations History panel for the required conversation session.



2. In the Labels window, scroll down and click + Add.
3. Enter the label name, and click the confirm icon.

Note: You can select a different label color by clicking the color icon and using the color palette.



4. Click Save.

Manage a Custom Label

Edit

You can edit or delete an existing custom label while adding a new label.

To edit a label, follow the steps below:

1. Click the + Label button on the Conversations History panel for the required conversation session.
2. In the Labels window, click the edit icon for the label you want to modify.
3. Make the changes and click the confirm icon.
4. Click Save.

Delete

To delete a label, follow the steps below:

1. Click the + Label button on the Conversations History panel for the required conversation session.
2. In the Labels window, click the delete icon for the label you want to remove.
3. Click Save.

Other Conversation Parameters

The Channel, Language, Date, and conversation duration are displayed at the top of the Conversation History analytics panel.



Conversation Summary

• User Messages: The number of messages sent by the user to the virtual assistant.
• Bot Messages: The number of messages sent by the bot/virtual assistant to the user.
• Intents Identified: The number of user intents identified by the virtual assistant during the conversation.
• Intent Unidentified: The number of utterances that did not result in any intent identification during the conversation.
• Tasks Completed: The number of tasks successfully completed by the virtual assistant.
• Failed Tasks: The number of tasks that failed during the conversation.

Conversation Events

Conversation Events indicate the sequence of occurrences during a conversation triggered by customer inputs and responses while interacting with a bot. Each event can be categorized into regular and error-based occurrences. When analyzing a conversation flow, the user can identify the event type based on the following icons that are displayed for the relevant event.



The list of events triggered during a conversation are displayed below.

Event	Description
Intent Found	Triggered when the virtual assistant understands the user’s intent. The intent name identified is displayed.
Intent Not Found	Triggered when the virtual assistant is unable to understand the user’s intent.
Agent Transfer	Triggered when a live agent transfer is initiated during the conversation.
Entity Retry	Triggered when the input provided by the user is not identified by the virtual assistant and a input retry request to the customer.
Confirmation Retry	Triggered when a confirmation request for a query or selection is generated from the virtual assistant to the customer.
On Connect	Triggered every time a customer invokes Web/Mobile SDK (a conversation is initiated on the channel).
Sentiment Event	Triggered when a customer’s sentiment is identified during the conversation. The user sentiment type identified is displayed.
Welcome	Triggered on receiving a message from user when channel specific event is not configured.
Welcome Event Telegram	Triggered on receiving a welcome event from Telegram.
Welcome Event Facebook	Triggered on receiving a welcome event from Facebook Messenger.
Welcome Event Telephone	Triggered on receiving a Telephone Call from any voice channel.
End	Triggered on the closing of a conversation.
Debug Log	Triggered when a debug log script runs for a script failure or service failure event.
Script Failure	Triggered when a Script node failure occurs.
Service Failure	Triggered when a service node failure occurs.
RCS Opt In	Triggered on receiving Opt-In from the user for Rich Communication Services.
RCS Opt Out	Triggered on receiving Opt-out from the user for Rich Communication Services.
User Message	Triggered when a customer sends a message to the virtual assistant on the channel.

Enriched Chat Transcript

The Enriched Chat Transcript feature provides a more detailed view of the conversation, with all the events associated with each message displayed in the chat transcript slider. This can be useful for understanding the conversation at a granular level and identifying any issues that may have arisen. It can also help train the virtual assistant and improve its performance.

To view the Enriched Chat Transcript, enable the Show Events option (default setting) in the Chat History panel.

The Event labels that appear under Conversation Events are displayed against each chat text in the chat transcript slider. Click here to learn more about Conversational Events.

Feedback User Type Label

For conversations with the customer response to a feedback survey, the feedback user type label is added automatically based on the analytics data. This label indicates if a customer is a promoter or a detractor.

Hovering the mouse on the label displays the following data:

• Number of feedback responses collected.
• Type of feedback survey.
• Customer type (result generated internally after the feedback is submitted).
• Feedback score.

User Details

To view the details of the agent that handled a conversation session, follow the steps below:

1. Click either Conversation Summary or Conversation Event on the dashboard.
2. On the landing page, click the User Details tab to view the following information:
   • User ID: The unique ID assigned to the agent.
   • Channel Data: The conversational channel where the interaction occurred.
   • Total Conversation Sessions: The total number of conversations the agent has handled for the selected period.
   • Sessions in the Last 30 Days: The number of conversations the agent handled in the date period that the used has selected within the last 30 days.
   • Last Interaction Date: The date of the agent’s most recent customer interaction.
   • User Meta tags: Used to identify important keywords in the conversation.

   

Chat History

The chat conversation flow between the virtual assistant and the customer is displayed in the Chat History panel, including events, actions, input requests, queries, and intent-based responses. Any failed tasks or exceptions that occurred during the conversation will also be displayed here.

The Enriched Chat Transcript is visible on the Chat History panel.

To view Chat History, follow these steps:

• Click an entry under Conversation Summary or an event under Conversation Events on the Conversation History dashboard.
• The Chat History window displays the conversation flow. In addition, the following information is displayed on the side pane:
  • Date Filter dropdown to select the period of Chat History.
  • The date-wise information summary on the following:
    • Conversation channel.
    • Chat initiation event.
    • Chat Duration
    • Chat Start time and End time.
    • Conversation summary of the following metrics:
      • User messages
      • Bot messages
      • Intents identified
      • Intents Unidentified
      • Tasks completed
      • Failed Tasks
    • Language (of interaction)
    • Event labels tagged to the conversation history.

    

Message Tags

Message Tags help identify and categorize messages in the chat transcript. They add clarity and context to the conversation and can be reused across related conversations. System-defined message tags are provided by default, but users can create custom tags by defining a key-value pair. The key is used as an identifier for the specific message, and the value is the expected customer response to that message.

To add a custom message tag, follow the steps below:

1. Hover over the message transcript in the conversation where you wish to add the message tag.
2. Click +Message Tag.



3. In the Add Message Tag window, enter the Key and Value in their respective text fields.



4. Click Save.

The new message tag is displayed below the message in the transcript slider.

The Feedback Survey feature allows customers to provide feedback on their conversations with the virtual assistant.

If enabled, the customer’s response to the survey will be displayed in the chat transcript under Chat History.

The system captures and displays a feedback event label “End of Conversation: Survey Type” on the chat transcript, indicating the end of the conversation and the survey type.

The response will be mapped to a key-value pair for the conversation timeline, which can be useful for analyzing the conversation and customer experience at a granular level. This information is useful to improve the virtual assistant’s performance and enhance the customer experience.

The Conversation Summary displays the following information on the Feedback event:

1. The Start Time and End Time, along with the real-time counts for the following:

• User Messages
• Bot Messages
• Intents Identified
• Intents Unidentified
• Tasks Completed
• Failed Tasks

2. The Feedback Event is based on the event timeline with Key and Value.
3. The total number of feedback responses collected for the session and the following:

• Type of Feedback
• Feedback response
• Score