Insights

This section discusses various insights available with Quality Management.

Quality Management generates a range of analytical insights from each analyzed interaction, accessible through the web portal. These insights include conversational summaries, sentiment and engagement scores, keyword and topic tracking, speech event detection, and reporting tools that aggregate performance data across agents and time periods. Together, they provide supervisors and administrators with a structured basis for evaluating contact center activity.

Interactions

The Interactions menu is the central location to review conversations and insights

Supervisors and administrators can centrally review all conversations analyzed by Quality Management through the Interactions sub-menu on the web portal.

Heads Up

You must be an authorized user to view interactions.

On the Interactions sub-menu, click the blue Interaction ID in any row to open a detailed view of that interaction. From there, users can play the recording, read the transcript, and view the associated analytics, comments, and details. After selecting an Interaction ID, Quality Management's conversational analysis is displayed on-screen.

The Analytics tab provides a conversational summary, including an engagement and sentiment score, Indicators and Topics identified, and more

When viewing an interaction, the Analytics tab provides a conversational summary including an engagement score, sentiment score, and identified Indicators and Topics. Specific insights include the following.

• Engagement is a rated score based on various factors, including talk-time ratio, response time lag, and frequency of speaker changes.

• Sentiment is a rated score based on the entire conversation's analysis, primarily focused on the sentiment of the conversation. Higher scores indicate a more positive effect; scores around 50 indicate neutral sentiment.

• Call Outs are specific words or phrases noted as significant within a conversation, determined by an account's configured Indicators. This section also notes Good Questions identified by AI analysis.

• Topic Highlights are broad, large-scale themes identified within a conversation, determined by an account's configured Topics.

• Speech Events reflect the number of identified instances where conversational silence or crosstalk exceeded defined thresholds, and the number of times a call was put on hold.

Interaction search supports Boolean operators and expression-based filtering

Supervisors can search interactions using expression-based queries with the following operators: AND, OR, NOT, quoted phrases, and wildcards. This refined search capability helps locate specific interactions without manually scrolling through the full list.

The interaction list can be sorted and filtered by Queue and Channel

Interactions can be sorted in ascending or descending order by queue assignment. Users can also filter the list to narrow results by specific queues or communication channels (voice, video, SMS, messaging, email).

Interactions can be exported to CSV for external analysis

Admins and supervisors can export the interaction list to a CSV file. The exported data reflects any filters currently applied (date range, channel, agent, etc.), making it easy to analyze large volumes of interaction data in external tools or share with stakeholders who do not have Quality Management access.

Relative date filters and configurable page sizes improve navigation efficiency

Users can filter interactions using "Today" and "Yesterday" shortcuts in addition to manual date ranges. The page size can be set to display 20, 50, or 100 interactions per page (default: 20), allowing faster review of large interaction volumes.

A filter shows only evaluated interactions on the My Interactions tab

Users can apply a filter on the My Interactions tab to display only interactions that have already received an evaluation, including both manual and automated (Auto QM) evaluations. The filter provides quick access to associated answers and total scores.

Transcripts can be translated into any supported language

Quality Management users can request a translation of the generated transcript, and conversation summary from the original language into any supported language. Translated versions can be saved for other users who have access to the interaction, supporting multilingual review teams.

Sentiment Analysis

Sentiment scores are derived from AI analysis of conversation transcripts

Sentiment scores do not account for vocal tone or volume — only transcript content. Higher scores indicate positivity; a score of 50 is neutral. Users can interact with sentiment graphs to navigate to relevant sections of the transcript.

AI-generated explanations accompany numeric sentiment scores

Alongside the numeric sentiment score, Quality Management provides a concise, AI-generated explanation (one to two sentences) describing why that rating was assigned. These explanations are based on transcript content and are disabled automatically if sentiment analysis is turned off at the account level.

Account owners and admins can disable sentiment analysis entirely

When sentiment analysis is disabled, all sentiment scores are hidden, the sentiment trend chart is removed from the analytics view, and the sentiment score column is removed from the interactions list and speaker metrics pages. This gives organizations flexibility to opt out of sentiment measurement if preferred or required.

Indicators (Call Outs)

Indicators (Call Outs) are customizable keywords and phrases tracked within conversations

Indicators, identified under the Call Outs section, are customizable keywords or phrases highlighted within a conversation's analysis. Indicators can be used to capture critical moments of a conversation or track mentions of a specific competitor, feature, product, or phrase. Accounts use Indicators to identify specific elements of conversation worth reviewing or tracking.

For example, a business focused on reducing churn relative to competitors can create Indicators for their own products and those of competitors. When an Indicator is mentioned within a conversation, the business can quickly identify the context of the mention and address the consumer's need or concern accordingly.

Example

If a business specializes in candy, they can create Indicators for the various types of candy, like Gummy Candy, along with specific examples of those candies, like gummy worms or gummy bears.

The following images provide an example of Indicators noted in a conversation’s analysis. Within this image, the Indicators Hard Candy and Gummy Candy are noted, because Indicator-specific keywords or phrases, like gummy bears and gummy worms were noted in the conversation’s transcript. After clicking an Indicator (for example, gummy candy), the transcript will reflect sections in the conversation where the Indicators were noted.

Indicators can be organized into custom categories

Accounts can create multiple custom categories to group and track Indicators mentioned within conversations. This provides flexibility to create classification systems suited to each account's needs, environment, or sector.

Example

If your business specializes in chocolate candy, you could create a Competitors category, with an Indicator for each of your primary competitors’ candy types, like hard candy or gummy candy. Using Quality Management, you can create an Indicator for Hard Candy, with the following keywords:

• lollipops

• candy canes

• butterscotch

You could also create a second Indicator for Gummy Candy, with the following keywords:

• gummy worms

• gummy bears

• fruit slices

When combined, account administrators can review which Indicators are most commonly mentioned, along with which of their competitors is mentioned, in addition to the specific product that is being mentioned. This can provide key context to a consumer’s sentiment and if they are interested in a competitor’s product(s).

The following image provides an example of the previous example’s Indicators and their associated keywords grouped under the common Competitors category.

Indicators match against the spoken language of an interaction

Indicators match against the spoken language of an interaction and its associated transcript. Indicators should be developed in the language or languages expected in the business environment.

Indicators can be retroactively applied to historical conversations

After adding new Indicators to an account, interactions will retroactively reflect the new Indicators upon analysis, helping businesses quantify common themes over time as trends become clear.

Accounts are allowed up to 250 Indicators

As of the date of this document’s publication, accounts are currently allowed up to 250 unique Indicators.

Indicators can be set up to be triggered by agents, consumers, or both

When creating an Indicator, account admins can configure if the Indicator will be noted in a conversation’s analysis based on who mentioned the keyword or phrase in the conversation.

For example, a contact center agent might regularly reference a product or set of features, which may skew analytics for Indicators mentioned. However, these data points may prove valuable if a consumer mentions these products or features, indicating their sentiment or business needs. Alternatively, an account may choose to always highlight indicator mentions, by either an agent or consumer, for quality assessment and complete coverage.

Indicator analytics show mention frequency, trends, and transcript-level context

Admins and supervisors can view the frequency and percentage of interactions containing each configured indicator, analyze trends over time, and filter data by account or user. Drilling into a specific keyword surfaces transcript snippets showing the context in which the phrase appeared.

AI can generate keyword and phrase suggestions when creating a new indicator

When creating an indicator, administrators can provide the indicator name and an optional description to receive AI-generated suggestions for relevant keywords and phrases. These suggestions can be reviewed, edited, and approved before the indicator is saved.

Indicators can be created and managed in bulk via CSV import and export

Admins can import a CSV file to create or edit multiple indicator categories and indicators at once. They can also export the complete indicator setup — including names, descriptions, categories, and keyword phrases — to a CSV for offline editing or team collaboration.

Automation rules can be configured to detect when an indicator is NOT mentioned

In addition to detecting when an indicator phrase is spoken, supervisors can create automation rules that flag interactions where a specific indicator or required phrase was absent. This is particularly useful for compliance use cases where certain disclosures or scripts are required on every call.

Subscriptions

Supervisors and administrators can subscribe to Indicator mention notifications

Quality Management supervisors and administrators can subscribe to Indicator mention notifications via email to track or monitor specific themes or trends. When subscribing, the user can choose to receive subscription notifications on a daily, weekly, monthly, or continuous (i.e., as-it-happens), basis.

For instance, a contact center manager may subscribe to an Indicator for escalation requests that track the frequencies of consumers requesting to speak or escalate a situation to a manager.

Subscriptions can be specific to a keyword, phrase, or the entire Indicator

Subscriptions can be scoped to a specific keyword, phrase, or an entire Indicator. For example, if the Indicator "Cancel Service" contains the keywords cancel, terminate, discontinue, end contract, and termination, a user can subscribe to all keywords or to specific phrases only.

Topics

Topics track broad, predefined themes within a conversation and provide insight into consumer concerns. Rather than tracking themes through specific keyword matching — as Indicators do — Quality Management uses AI and ML algorithms to identify questions or statements relevant to a predefined theme. Topics can surface relevant conversation segments even when the exact language varies.

Quality Management includes seven default topics and supports up to 10 additional custom topics

Quality Management includes seven default Topics that it automatically tracks within conversations: Pricing, Legal, Privacy, Security, User Requirements, Hardware, and Licensing. Accounts can configure up to 10 additional custom Topics; account admins must provide a variety of guided sentences through which the Topic's theme can be understood by the AI and ML algorithms.

The following image contains an example of the Privacy Topic and its foundational guiding sentences.

Topics match exclusively against the transcript of an interaction

Unlike Indicators, which identify spoken words within the spoken language, Topics are processed through an AI and ML engine. Conversations in supported languages other than English are first transcribed into English before Topics matching occurs. When a Topic is created in a supported language other than English, it is translated to English for storage and matched against the translated transcript accordingly.

AI-powered topic detection works across all channel types

Topic detection has been enhanced with AI-powered identification that automatically highlights key topics within interactions. The system identifies topics across voice, video, SMS, and chat interactions. Supervisors can view detected topics from both the interaction list and the interaction detail page, and can see the specific segments of each interaction where those topics were detected.

A "custom topics only" toggle simplifies topic mention analytics

When analyzing topic mentions, supervisors can enable a single-click filter to display only custom topics, hiding the seven default topics from the analytics view.

AI-detected topics are included in CSV exports of interaction data

When interaction data is exported to CSV, a dedicated Topics column includes the AI-detected topics for each interaction, enabling users to spot trends and patterns without reviewing each interaction individually.

Speech Events

Speech Events identify moments of silence, crosstalk, and hold

Speech Events help supervisors identify moments in an interaction where silence or crosstalk exceeded an account's defined threshold settings, as well as the number of times a call was placed on hold. This supports supervisors in confirming that interactions meet organizational standards and in identifying where in an interaction these events occurred.

Speech Events are outlined and labeled below the interaction’s media player

When viewing an interaction, supervisors can see an outline and chronology of the conversation's Speech Events below the media player. This view shows who spoke, when they spoke, and the type and duration of each triggered event.

Hold events show call duration and are visualized in the media player

For each interaction, users can view the total number of hold events, each event's individual duration, and a visual indicator of hold events directly in the media player timeline. This helps identify how often and how long callers were placed on hold, supporting quality review and coaching decisions.

The silence event threshold can be increased to up to 60 seconds

Administrators can raise the silence detection threshold to up to 60 seconds, preventing routine pauses from generating silence events and reducing noise in the analytics view.

Agent Summary and Speaker Metrics

Each conversation includes an Agent Summary, providing details about how the agent interacts with the consumer, like the talk-to-listen ratio, filler words used, and more

Each conversation analyzed by Quality Management includes an Agent Summary providing insight into how agents interact with consumers. The Agent Summary includes the user’s talk-to-listen ratio, the longest uninterrupted spiel, the number of filler words used per minute, average talking speed, and the average time the agent waits to respond. With these metrics, managers can create coaching opportunities to improve agent performance over time.

Aggregated speaker metrics analytics visualize performance trends over time

Supervisors can view aggregated speaker metrics for individual agents or across teams, with trend visualization over a selected time range. Data can be grouped by agent and filtered by date range, queue, channel, language, and other dimensions. Metrics include talking speed, talk-listen ratio, filler words, longest monologue, patience, sentiment score, and engagement score.

Four new duration-based metrics are available for voice and video interactions

The following metrics are now available in interaction details, automated scorecards, evaluation goals, and via API: participant talk duration, total hold duration, total silence duration, and total talk-over duration.

Administrators can configure recommended score ranges for speech analytics metrics

Admins can customize what counts as "positive," "neutral," or "needs improvement" for sentiment score, engagement score, talk speed, filler words, talk-listen ratio, longest monologue, and patience. Visuals and tooltips across Quality Management analytics update to reflect the configured thresholds.

Reporting

Quality Management includes enhanced reporting capabilities

The following reports and dashboards are available within Quality Management.

• The Indicator Mentions Report is a dynamic report displaying the frequency and percentage of interactions containing configured Indicators, phrases, or keywords, with trend analysis over time.

• The Speaker Metrics Report provides an overview of individual agent performance over time to track changes from coaching and identify trends.

• The Evaluation Scores Report provides a summary of agent evaluations over specified time frames to identify top and bottom performers.

• The Response Breakdown by Question Report is a visual representation of scorecard scores in pie or bar chart formats.

• The Evaluator Performance Historical Report details evaluation completion by supervisors, including completed evaluations, average scores, and evaluation dates.

• The Trending Topics Dashboard provides real-time insights into topics customers raise when contacting the call center, enabling supervisors to identify emerging issues, spikes in discussion topics, and trends in customer sentiment. Supervisors can apply multi-dimensional filters, including team, queue, and channel, to focus the dashboard on specific segments of activity. Frequently used filter combinations can be saved as filter groups and reapplied with a single selection. Note: The Trending Topics Dashboard requires an Advanced QM license.

• The Topic Mentions Report provides detailed insights into the broader themes and subjects discussed during customer interactions. Supervisors can filter Topic Mention analytics by Team to associate topic trends with specific teams. Frequently used filter combinations can be saved as filter groups and reapplied with a single selection; filter groups are scoped to the user who created them.

The Speaker Metrics Report tracks agent performance over time

The Speaker Metrics report provides an overview of an individual agent's performance over time. This enables supervisors to track agent performance — including changes as a result of coaching and feedback — and identify trends such as changes in speech events, talking speed, or talk-to-listen ratio.

The Subscriptions section allows supervisors to automatically subscribe to recurring reports on a daily, weekly, or monthly basis for the Evaluation Results by Agent and Evaluator Performance reports.

The Indicator Mentions report provides insight into Indicator frequency and trends

The Indicator Mentions report displays the frequency and percentage of interactions containing configured Indicators, phrases, or keywords, and supports trend analysis over time. Supervisors can use this report to monitor interactions and identify issues and trends.

After reviewing an Indicator's aggregate analytics, additional detail is available in line or bubble view. Supervisors can drill further into Indicator analysis to see specific conversations and moments in the transcript where the Indicator was mentioned, including the ability to filter data by account or user and review transcript snippets for individual keywords.

For example, the following image shows additional information for the Churn risk Indicator.

From there, supervisors can drill further into Indicator analysis and see specific conversations and moments in the transcript where the Indicator was mentioned. This includes the ability to filter data by account or user and drill down into specific keywords to see transcript snippets for context. For example, the following image shows mentions of the keyword Cancel, which is configured as a part of the Churn risk Indicator.

Analytics data can be exported to CSV or PDF

Users with the Analytics and Reports export permission can export data from the Answer Distribution Analytics page to CSV or PDF. Exported data includes question details, response metrics, section mappings, and scorecard information, and respects any filters applied prior to export.

Indicator mentions analytics can be exported to CSV

Supervisors with the Analytics and Reports export permission can export Indicator Mentions analytics data to CSV. Two export formats are available: a Summary export providing mention frequency and percentage data per indicator, and a Detailed export providing a row-level breakdown with agent, team, and queue attribution for each mention. Any filters applied to the Indicator Mentions view are respected in the exported file.

Agent email addresses are visible alongside display names in analytics

Supervisors can view agent email addresses alongside display names in evaluation results, evaluator performance reports, and speaker metrics — providing a secondary identifier when multiple agents share similar names.

Frequently used filter combinations can be saved for quick reuse

Supervisors can save commonly used filter combinations in Quality Management analytics to quickly return to a configured view without manually re-selecting filters each session. Supervisors can save commonly used filter combinations in Quality Management analytics to quickly return to a configured view without manually re-selecting filters each session.

Analytics data, topics, indicators, and evaluations are available via API

REST API endpoints are available to retrieve: (1) conversation topics and performance indicators from interactions; and (2) a list of Quality Management evaluations, filterable by acknowledgement status, agent ID, evaluator ID, evaluation status, and interaction ID. Evaluation data is returned in JSON format for use with external reporting systems.

Topic mentions analytics data can be exported to CSV

Users with the Analytics and Reports export permission can download a complete CSV of topic mention rate data from the Topic Mentions page. The exported file includes all topics matching the currently active filters and contains the following columns:

• Topic name

• Topic type

• Mention rate

• Interaction count

The file header captures the active filter context at the time of export, including: Time period, Selected team(s), Selected queue(s), Selected channel(s), Topic type, Total interactions, and Searched specific topics. The Export button enters a loading state during file generation and displays an error state if generation fails.

Team Interactions permission now grants access to analytics reports and automation configuration scoped to assigned teams

Users with the Team Interactions permission can view Quality Management analytics reports and configure automation rules scoped to their assigned teams without requiring the All Interactions permission. Report data and automation rules only reflect the interactions for the user's assigned team or teams.

Note This change doesn’t impact users with the All Interactions permission. They’ll continue to see data across all teams.