Conversation Intelligence

Overview

The Conversation Intelligence dashboard displays post-interaction analytics and provides insights following customer interactions. It gathers and presents data from those interactions, helping admins and supervisors understand agent performance, customer experience, and key interaction details.

The key sections of this dashboard include:

• Contact Center Efficiency: Key Performance Indicators (Average Speed to Answer, Abandonment Rate, CSAT, and Transfer Rate).

• Agent Efficacy: Empathy Score, CSAT, Crutch Word Score, Agent Performance Monitor, and Agent Occupancy.

• Customer Experience: Avg. Wait Time, Net Promoter Score (NPS), Churn Risk, Sentiment Score, Sentiment Monitor, and Customer Churn Monitor.

• Insights Mining: Topic vs. Sentiment Bubble, Keyword Cloud & Emotions.

• Interaction Details: Date/Time of the session, Call ID, Agent Name/ID, Call Reason/Intent, Sentiment Score, and Dispositions.

Access Conversation Intelligence

Navigate to Quality AI > ANALYZE > Conversation Intelligence.

Filters

You can use the filters to customize the Conversation Intelligence tab.

Create a Filter

Steps to create a filter:

1. Select the Filters.
   

2. Select + Add New Filter.
   

3. On the New Filter window, choose the relevant Queues or Agents, and select Apply.
   

   Note

   Select Apply saves filters to the Unsaved Filter category for Queues or Agents, letting you review them before saving permanently.

4. Under the Save Filter section, enter the Filter Name.

5. Enable the toggle switch Make this the default view for the new filter added or save filter.

6. Select Save & Apply. The system displays a confirmation message after it saves the filter.

Saved Filters

You can view the saved filters by selecting the Filters tab.

Hovering over a saved filter displays the following options:

• Duplicate
• Mark as Default
• Delete
• Edit
  

Duplicate Filter

Steps to duplicate a filter:

1. Select Duplicate to copy a filter.
   

2. Enter a name in the Filter Name field for the new filter, and select SAVE.
   

3. The system displays a confirmation message and creates a copy of the filter.
   

Mark as Default

Select the Mark as Default to set the filter as the default filter.

The system displays a confirmation message when the filter sets as the default filter.

Delete Filter

Steps to delete a filter:

1. Select the Delete.
   

2. On the delete confirmation pop-up window, select Delete.
   

   The system displays a confirmation message and deletes the filter.

Edit Filter

Steps to edit a filter:

1. Select the Edit.
   

2. Make the necessary changes, and select Save & Apply.
   

Clear Filters

This filter displays only the analytics data for the current day's agent performance, customer experience, and interaction details. When applied, it removes all previously selected date ranges and focuses solely on today's data.

Date and Time Range Selection

Lets you filter data by date and time range. By default, the current day data displays. Select a specified date and time range, and select Apply.

Compare Functionality

The Conversation Intelligence dashboard lets you view a comparison of metrics between a selected date range and the previous date range. By default, the system enables the compare toggle.

Spike and dip indicators for playbook adherence link to this toggle and appear across stages and steps only when enabled. The system includes these indicators for all fields. An increase or decrease can have different meanings depending on the metrics, resulting in either red badges (negative) or green badges (positive).

For an example, if you select the last seven days as the date range, the system shows a spike or dip for all metrics, such as Average Speed to Answer, indicating the percentage change for the period (June 5 to June 12) compared with the same duration in the previous period (May 29 to June 5).

Channels

You can filter data by channels (All, Voice, Chat, and Email). This filter applies to the entire dashboard except for agent occupancy, and the since occupancy per channel is not tracked.

Contact Center Efficiency

Key Performance Indicators

Efficient contact centers are the backbone of excellent support. Managers use key performance indicators (KPIs) to ensure teams maximize productivity and efficiency. The KPIs in a contact center environment are measurable metrics used to assess the effectiveness and efficiency of customer service operations. They include:

Average Speed to Answer (Voice, Chat, Email channels): Average Speed to Answer (ASA) displays the average time an agent takes to answer inbound calls, starting when callers join the queue.

Abandonment Rate (Voice and Chat channels): The Abandonment Rate refers to the total number of customers who disconnect their calls while in the queue before reaching an agent.

CSAT (Voice and Chat channels): Customer Satisfaction (CSAT) scores indicate the level of customer satisfaction with Support services. This can be calculated from CSAT survey scores.

Transfer Rate (Voice, Chat, and Email channels): Transfer Rate measures the percentage of customer interactions that were transferred from one agent or queue to another to resolve the issue.

Default Zones for KPIs

• For Average Speed to Answer:
  • Voice Channel – Green up to 28 seconds, Yellow 28 to 40 seconds, Red 40 seconds and above.
  • Chat Channel – Green up to 35 seconds, Yellow 35 seconds to 50 seconds, Red 50 seconds and above.
• For Transfer Rate: Voice and Chat Channels – Green up to 10%, and Red beyond 10%.
• For Abandonment Rate: Voice and Chat Channels – Green up to 6%, and Red beyond 6%.
• For CSAT: Show the actuals as it is with the respective weight-ages of interactions.
  • CSAT is shown as a score in the range of 1-10 (1 being the lowest and 10 being the highest).
  • Green 8 and above, Yellow 6 to 8, and Red 1 to 6.

The average of the KPIs for a time interval is shown beside the line bar, along with a percentage increase or decrease in comparison with the period selected.

Hovering over the color zones shows the percentage of calls and the actual number of calls for the represented percentage (volume).

Agent Efficacy

Empathy Score

Empathy is a crucial necessity for having a quality customer experience. The empathy detection module analyzes both customer and agent utterances and detects if there is a need for empathy. This is done by detecting the sentiment of customer utterance and classifying it either as ’empathy-seeking’ or ‘non-empathy-seeking’.

The empathy module analyzes an agent’s utterance and classifies whether or not the agent is empathetic.

CSAT

Customer Satisfaction (CSAT) is a widely used metric for measuring and evaluating customer satisfaction. This widget displays the average of customers' CSAT scores on a scale of 1 to 5.

Crutch Word Score

This measures how frequently an agent uses crutch words during their speech. These words or phrases are used as fillers by the agents that give them more time to think. Common crutch words are “um,” “uh,” “like,” “you know,” “so,” and “basically.” The crutch word score is calculated for agent utterances and supports multilingual conversations, enhancing detection by identifying subtle crutch words. Learn more.

Agent Performance Monitor

The admins and supervisors must completely understand customer interactions and their influence on customer satisfaction. Agent Performance Monitor provides a visual representation of the relationship among different metrics, including empathy score, crutch word score, CSAT score, and sentiment score. You can make informed, data-driven decisions by identifying the patterns to enhance the overall customer experience.

The agent performance metrics are on the Y-axis. You can select individual or multiple metrics to view and analyze. CSAT and sentiment score are the customer experience-based metrics on the X-axis.

Agent Occupancy and Agent Status

Agent occupancy refers to the time agents actively handle customer interactions or perform work-related tasks like After Call Work (ACW). It represents the amount of time agents spend on productive activities compared to their total available working time.

Agent Occupancy (%) = (Total talk or chat time + Total ACW time) / (Total logged-in time) * 100

Total talk or chat time: The aggregate duration during which agents actively participate in customer interactions, including inbound and outbound calls or chats.

Total ACW time: The cumulative duration agents spend in post-call activities, including ACW, note-taking, updating customer records, or completing essential tasks linked to the interaction.

Total logged-in time: The total duration agents remain logged in to SmartAssist, available to handle customer interactions.

Agent statuses include available, busy, away, “break”, and other custom codes the administrator configures. This information is represented in a pie chart, enabling admins and supervisors to assess agents' distribution across various statuses quickly. This section also helps understand agent occupancy and optimize agent utilization.

Script and Playbook Adherence

Agents follow predefined scripts when handling customer interactions. These scripts are part of either the static playbooks configured in Agent AI or business process-specific scripts configured in the dynamic playbooks in Agent AI. Supervisors can track agents’ adherence to the configured scripts.

The adherence is presented as a bar graph, with each attribute represented by individual bars indicating the compliance percentage for the selected period.

The compliance percentage for each attribute is displayed beside the graph, along with an indication for an increase or decrease in compliance percentage during the selected period compared to the same time range. For example, if a supervisor views the compliance scores for the last week, the percentage increase or decrease is indicated for the previous week. Similarly, if the supervisor selects a one-month time frame, the comparison period is one month.

When the playbook steps are the same across all the different playbook configurations, they appear as the default set of parameters for measuring adherence.

When no Agent AI playbooks are configured, adherence is displayed for the following scripts (Conversation Etiquettes) by default:

• Greeting: For example, Hello, My name is John Doe, and I am your customer support executive. How may I help you today?

• Branding: For example, Thank you for contacting XYZ.

• Privacy Policy: For example, This call is being recorded for quality and training purposes.

• Hold Etiquette (Voice calls): For example, May I place you on hold for a few minutes while I pull up some information?

• Customer Verification: For example, Can I know your date of birth?

• Proper Sign Off: For example, Thank you for calling XYZ today. It was a pleasure to assist you. Have a great day!

  

  Supervisors can choose specific playbooks from the dropdown menu other than the default options. In that case, adherence to the specific parameters/steps in that playbook is displayed.

  

  Administrators can configure the attributes for agent playbook adherence. Learn more.

  Note

  The Playbook Adherence tab in Conversation Intelligence is available only when Playbooks are configured in Agent AI.

Customer Experience

You can monitor and review historical data to assess customer experience. This in-depth analysis of the metrics results in improved customer experience.

The following parameters can be monitored:

• All Languages: Enables the multi-selection of languages from the All Languages drop-down menu, reflecting the languages chosen in the evaluation form. By default, all languages are pre-selected. Only the metrics corresponding to the selected languages are shown under this filter. Once the language filter is applied, the relevant language-specific interaction data is displayed in the Sentiment Monitor and Customer Churn Monitor widgets.
• Average Wait Time: The average wait time is the total wait time for all customers in a specific period (for example, an hour) divided by the total number of customers served during that period.
• NPS Score: The Net Promoter Score (NPS) measures customer loyalty and satisfaction. It is determined through a survey question that asks customers to rate, on a scale from 0 to 10, how likely they are to recommend the products or services to others.
• Churn Risk: Churn risk refers to the rate at which customers stop using the services or products provided by the contact center and switch to a competitor or discontinue their engagement altogether. It indicates the loss of customers over a specific period.
• Sentiment Score: The sentiment of each customer utterance is assessed in real-time at the keyword/phrase/sub-topic level and stored. Once the call concludes, all the customer sentiments recorded are averaged to compute a sentiment score for the entire call, which is then normalized within the range of 1 to 10. It is a numerical representation of the sentiment or emotion expressed during the interaction, such as a comment. The goal is to determine the overall sentiment or attitude conveyed in the interaction.

Sentiment Monitor

The Sentiment Monitor module provides a score range from 1 to 10 for each interaction and indicates polarity as Positive, Neutral, or Negative. It determines the likely emotion involved based on the average sentiment score, such as Happy, Satisfied, or Disappointed.

The sentiment monitor graph shows the distribution of customer sentiment across different intents, enabling an admin to understand customer satisfaction and sentiment levels related to specific intents or topics of conversation.

The sentiment monitor graph is a bar chart where each bar represents an intent or topic. Each bar represents the associated sentiment across all interactions with that intent, categorizing the sentiment weightage as follows:

• Positive Sentiment: Green color
• Neutral Sentiment: Yellow color
• Negative Sentiment: Red color

Customer Churn Monitor

The Customer Churn Monitor shows customer churn risk in a pie chart. It compares the churn percentage with total calls and any escalations in the same period. Hovering over a section shows its value, and selecting a section opens the calls linked to that escalation or churn category. The dashboard displays the following details:

• No Customer Churn/Escalation: Number of interactions with no customer churn or escalation.

• Customer Churn: Count of interactions where there has been a customer churn.

• Escalation: Number of interactions where the customer has requested assistance from a supervisor or a senior.
  

Insights Mining

Insights Mining displays the top 30 topics by volume and lets you identify the topics associated with volume, sentiment, keywords and emotions for the selected date range and channel.

By default, this widget displays a blank screen without any data. Based on the date range and topic selected from the calendar, it pulls the required sentiment scores for interactions in bubbles.

All Languages

This lets you select multiple languages from the All Languages drop-down menu that match the languages in the evaluation form. By default, the system selects all languages. The filter shows only the metrics configured for the languages you choose.

You can filter data by channel (All, Voice, Chat, Email). This filter affects the entire dashboard, except for agent occupancy, since occupancy per channel is not tracked.

When you select any language or channel filter, the system updates the interaction sentiment scores and bubble-plot visuals in Topics and Keyword Cloud & Emotions across all widgets.

Topics

This shows graded sentiment scores for interactions on a scale of 1 to 10. Each topic appears as a bubble that reflects the volume of positive, neutral, or negative sentiment. The widget also shows the top five emotions, including both positive and negative ones.

When you hover over a topic, the tooltip shows the average emotion label (for example, angry or frustrated) instead of the topic name.

Bubble Color & Sentiment Logic

The bubble plotting visual representation helps you identify sentiment distribution and volume for each topic. Hovering over a bubble displays a tooltip with sentiment distribution, total interactions, and average sentiment score.

Sentiment Color

• Green (Positive): Indicates positive sentiment distribution and the Sentiment score is between 6 and 10.

• Grey (Neutral): Indicates neutral sentiment distribution and the Sentiment score is between 4 and 6.

• Red (Negative): Indicates negative sentiment distribution and the Sentiment score is between 1 and 4.

• No. of Interactions: Total count of interactions for that topic.

• Sentiment Score: Overall sentiment score for the topic.

The bubble size reflects the interaction volume for a topic. The system calculates this size dynamically within the selected date range and scales it to show the topic’s volume relative to the minimum and maximum values. By default, all the three selected sentiment colors, give you a combined view of all topics across sentiments. You can select a specific sentiment color to view the most relevant topics in that category.

The color logic for sentiment:

• 1<=x<4: Negative (Red)

• 4<=x<6: Neutral (Grey)

• 6<=x<=10: Positive (Green)

  Note

  By default, only the top most significant 30 topics (based on interaction volume) are displayed on the widget.

For example, the Payment bubble indicates sentiment distribution as Positive 68% (green), Neutral 10% (gray), and Negative 22% (red).

These percentages are represented by the bubble’s circumference. Hovering over the Payment bubble shows the average emotion index, such as Flexible, Accurate, Fast, and so on (associated with the average sentiment score).

Keyword Cloud & Emotions for All

This displays relevant keywords for a selected Topics from the previous widget. It excludes general stop words and common words.

Keyword Search

Supervisors can search specific keywords across interactions related to a chosen topic from the topic bubbles using the Keyword Search option.

You can select a single or all topics from the dropdown list to search for a keyword. The total mentions of the keyword and the total number of unique interactions where the keyword is displayed when hovering over the keyword.

If a specific topic bubble is selected, the topic dropdown list automatically chooses the relevant topic. The searched keyword appears in the selected topic bubble or the selected topic from the dropdown list.

When “all topics” are selected from the dropdown list, the topic bubble is deselected and it displays the keywords and interaction details across all topics.

Selecting a specific topic from the dropdown list selects the relevant topic bubble, which shows only the pertinent keywords for that topic.

If a keyword is present in multiple topics, the keyword count and interaction mention the specific topic’s relevance.

When searching for a keyword, only that specific keyword appears in the keyword cloud. Relevant semantic keywords, if available, are displayed in the keyword cloud.

The interaction details refer to the main keyword searched, and if any semantic variations are present, the interaction details are shown after the complete set of interactions for the specific keyword is searched.

When a specific semantic variation keyword is selected in the keyword cloud, the interactions related to that keyword are displayed in the interaction details.

Interaction Details

The Interaction Details widget shows information based on default settings, which you can configure. When you select specific topics or keywords, the widget displays the relevant information; if you make no selection, it shows data based on the highest sentiment score or other settings you configure.

This section includes the following fields:

• Date/Time of the session
• Call ID
• Agent name/ID (you can select from a dropdown list of agent groups/agents)
• Description (text reference to the keyword cloud)
• Call Reason/Intent
• Sentiment Score (configurable from high to low)
• Dispositions

Send Feedback