OCP® Reporting User Guide

Overview

OCP® Reporting is an out-of-the-box reporting and analytics platform available for all conversations through OCP® applications. These involve Orchestrator Application dialogs, Flows, and miniApps innovations originating from calls or web chats.

OCP Reporting® provides rich, graphical dashboards with key performance metrics in order to pinpoint opportunities for continuously improving the Virtual Assistant and gaining a deeper insight into what customers are saying. The reports are unsupervised as they come directly from the system without any analysis from a human.

To access OCP® Reporting, go to Insights → Reporting in the side panel.

The OCP® Reporting is divided into the following tabs:

• Real-time report puts emphasis on near real-time data.

• Conversational Insights offers more analytical historical metrics.

• Voice Biometrics reporting displays the enrollments and verifications data.

Real-time report

The Real-time report tab shows the necessary information about the current status of your miniApp. These reports are created from a specific selection of existing batch metrics, targeted to ensure the user is informed about the current state of their system, without being overwhelmed with information.

To produce near real-time reports on top of the underlying data, the Data Warehouse infrastructure is used. Reporting metrics are broken down into 5-minute intervals and updated near real-time, with an additional watermark of 5 minutes. This means any data delay (due to external services) can be captured even with up to a five-minute delay. These metrics are publicly available through the Metrics API.

You can refresh the results at any time by clicking the Refresh button as shown below.

The default Time period for the data displayed is Last 120 minutes. However, by clicking on the Time period dropdown, you can select from predefined time period options ranging from Last 20 minutes to Last 48 hours.

Real-time report graphs

The following information is available in the Real-time report tab:

• Active Sessions

• Average Sessions & System Status

• Conversational Outcome (Unique Dialogs)

• Terminated VS Transferred to Agent

• No Input/No Match Events

• Total Tasks (miniApps)

• miniApp Success rate

Active Sessions

The Active Sessions report shows the number of distinct sessions the application received.

Sessions refer to either the number of unique IVR calls or chat sessions (depending on the application), no matter how many miniApps or Orchestrator Flows were triggered. For the session counter to increase, at least one miniApp must be triggered within OCP®.

Average Sessions & System Status

The Sessions panel of the right to the Active Sessions report shows the average number of sessions per minute that the application received for the last 2 hours (by default). If you change the time period in the Time period dropdown, the time frame of this panel will update accordingly.

The System Status panel displays the percentage of system and critical errors over total tasks of the last 2 hours (by default). If you change the time period in the Time period dropdown, the time frame of this panel will update accordingly.

Conversational Outcome (Unique Dialogs)

The Conversational Outcome (Unique Dialogs) report is applicable for Orchestrator applications and refers to the number of dialogs per Orchestrator application. A dialog is a unique conversation (call or chat) within an OCP® application.

The following metrics are available:

• Completed: Number of completed dialogs. Includes both terminated and transferred dialogs, as well as edge cases, when the dialog end is unknown. 

• Transferred: Total number of dialogs that got transferred to a live agent. 

• Terminated: Total number of dialogs that got terminated by the user or the system. 

Terminated VS Transferred to Agent

The Terminated VS Transferred to Agent report displays a number of dialog termination and transfer rates over total calls of the last 2 hours (by default). If you change the time period in the Time period dropdown, the time frame of this panel will update accordingly.

• Terminated: Total number or terminated calls, including both user and system termination. The percentage below the total number of terminated calls shows the termination rate over total call volume within the selected time frame.

• Transferred: Total number of transferred calls outside of IVR. The percentage below the total number of transferred calls shows the transfer rate over total call volume within the selected time frame.

No Input/No Match Events

The No Input/No Match Events report displays a total number of No Inputs and No Matches per miniApp. 

No Inputs and No Matches are applicable only for those miniApps that expect a user input. 

• No Inputs: Total number of steps with no input. A single miniApp may have multiple steps with no inputs. This is a per-step metric.

• No Matches: Total number of steps with no match. A single miniApp may have multiple steps with no matches. This is a per-step metric. 

Total Tasks (miniApps)

The Total Tasks (miniApps) report is applicable for OCP® miniApp applications. Each counter is increased after the compilation of each miniApp task. 

The following metrics are available:

• Total: The number of total invocations (tasks) originating from miniApps. This includes successful miniApps, miniApps that ended with system errors and critical errors, as well as miniApps that ended with any other failed reason like user termination or invalid input.

• Successful: The number of total successful miniApps invocations. 

• System Errors: The number of invocations ended with a system error or a timeout error. 

• Critical Error: The number of invocations ended with a critical error. A critical error may affect one or multiple Applications within the same call.

miniApp Success rate

To the right of the Total Tasks (miniApps) graph you can find a Success Rate graph for last 2 hours (by default). If you change the time period in the Time period dropdown, the time frame of this panel will update accordingly.

The Success Rate graph displays the percentage of successful invocations of miniApps (green) over the total miniApp invocations.

The Successful metric highlights the percentage of successful miniApp invocations over the total number of miniApp invocations.

Keep in mind unsuccessful miniApps are not only errors but cases where could be a business failure, early hang ups, etc.

Real-time report filters

The data results for real-time data are based on the filters set in the Filters menu. The search filter allows you to easily target specific applications or miniApps within OCP®.

To open the search filters, proceed as follows:

1. Click the Filters button, select the filters you want to pin and click the Apply button.

2. Each pinned filter has a dropdown list of available options. Fill in the search filter fields as per your needs.
   

   

   
   

• Group: The OCP group that the metrics refer to. Only one group may be selected each time.

• Name: The name of the miniApp for the miniApp Invocations report or Orchestrator application for the Dialogs report.

• Channel: The channel of the user system interaction (for example, Voice or Chat). Applicable for all reports. 

• Language: The selected language of the user system interaction (for example, English US).

• Region: OCP Environment Origin Region.

• Test Flag: Select true for fetching only test-flagged calls (or char sessions). To exclude test-flagged calls, select false. By default, all calls are reported. Applicable for all reports. 

• Type: The type of the miniApp shown in the reports. For Orchestrator applications, the type is always Flow. Not applicable for Session reports.

3. Click the Apply button to get the real-time data results.

4. Click the Clear All button to reset the filters. This will reset all search filter fields.

Viewing data in Full Screen mode

You can view any graph in the full-screen mode by clicking the Full Screen button next to each graph.

It is possible to narrow down the time frame displayed in the report viewed in the full-mode by clicking the blue dot at any end of the slider element and dragging it to the left or right.

You can close the Full Screen mode by clicking the Close button in the right bottom corner or the grey area outside the modal window.

Downloading data

You can download each report locally as a CSV file by clicking the Export Data button next to each graph.

When in the Full Screen mode, you can also download the report by clicking the Export button. If you change the time range using the slider element and then click the Export button, you can download data for the custom time period you selected.

Selecting metrics

You can select / deselect the metrics displayed in a report by interacting with checkboxes under each graph.

• Hover over a specific checkbox to view data related to its content only. Once you hover away from the checkbox, the report will display all metrics again.

• Click on the checkbox to make only this metric selected and displayed. You can click multiple checkboxes to display the needed metrics only. Deselecting all checkboxes returns the graph to the initial state where all the metrics are displayed.

Viewing metrics details

To view the metrics details, proceed as follows:

1. Hover over the graph to see the metric’s time.

2. Click on a dot with a specific time.

3. Click the Show Actual Dialogs button. You will be redirected to OCP® Monitor to review the dialog IDs corresponding to the reported metric.

Conversational Insights

OCP Conversational Insights® provides rich, graphical dashboards with key performance metrics and come directly from the system without any analysis from a human.

There are several dashboards to help you identify how your applications are performing and point out possible improvements:

• Total Sessions

• Avg Invocations per Session Daily

• Conversational Outcome

• Total Tasks (miniApps)

• Task Success Rate (miniApps)

• Agent Success Rate

Total Sessions

The Total Sessions interactive visualization shows the number of sessions that a customer is receiving for the selected period. Sessions refer to unique calls or chat sessions, regardless of how many applications were triggered in between.

Avg Invocations per Session Daily

The Avg Invocations per Session Daily graph connects sessions to miniApp invocations. Presents how many miniApps are triggered during an average call or chat session.

Conversational Outcome

The Conversational Outcome dashboard allows observing number of total dialogs per day.

The data is presented both in numbers and percentages.

• Completed - how many dialogs have been completed by the user or system. Includes both terminated and transferred dialogs, as well as edge cases, when the dialog end is unknown. 

• Terminated - how many dialogs have been terminated by the user or system.

• Transferred - how many dialogs have been transferred out of the system, usually to a live agent.

• Error - how many dialogs have been completed with an error.

• Other - dialogs which are not classified in any of the above cases. These are not necessary errors, but account for possible edge cases of dialogs for which there wasn’t enough information received to determine the exact termination type.

Total Tasks (miniApps)

A graphical representation of the total invocations during the selected period of time, as well as the percentage of the successful against the unsuccessful ones due to fail or error.

Task Success Rate (miniApps)

A graphical representation of the the miniApp results of all the invocations within the selected period. The Multi-line graph is customizable and is possible to select to view only the desired results. Click on the checkboxes to remove or add back a task category.

The Task Success Rate (miniApps) graph shows the success and failure rates of the selected miniApps. This is the task completion rate of each miniApp, which breaks down in the Overall miniApp Success Rate graph to:

• Successful Invocations - the percentage of successful invocations over the total number of invocations on a daily basis.

• Failed Invocations - the percentage of failed invocations over the total number of invocations on a daily basis.

• Terminated - the percentage of terminated by the user invocations over the total number of invocations on a daily basis.

• DTMF Escape Key - the total number of invocations within which the user opted out using the DTMF escape key over the total number of invocations on a daily basis.

• Errors - the total number of invocations with at least one system error over the total number of invocations on a per-day basis.

• Other - a final categorization that captures all invocations not falling in the above categories. These may include yet unclassified errors or edge cases.

Task Agents Reporting

Task Agents Reporting surfaces performance and outcome data for Agentic AI tasks executed within your dialog flows. Each time a task agent runs inside a conversation, its outcome is recorded and aggregated, following the same reporting logic already used for Self-Service and miniApps metrics.

The report answers a simple operational question: how many task agents ran, and how many of them succeeded versus failed.

You can find the Task Agents reports in the Conversational Insights tab, alongside the existing dashboards.

Total Tasks (Task Agents)

The Total Tasks (Task Agents) report shows the total number of task agent invocations over the selected period, broken down by outcome. Each counter increases after a task agent completes within a dialog.

The following metrics are available:

• Total: The total number of task agent invocations, regardless of outcome.

• Completed: The number of task agent invocations that completed successfully.

• Failed: The number of task agent invocations that ended in failure.

Task Agent Success Rate

The Task Agent Success Rate report displays the proportion of successful task agent invocations (completed) over the total number of invocations for the selected period. It mirrors the Task Success Rate (miniApps) report and is presented both in numbers and percentages.

• Completed: The percentage of task agents that completed successfully over the total number of invocations.

• Failed: The percentage of task agents that failed over the total number of invocations.

Filtering by Agentic App Name

A new Agentic App Name filter lets you narrow the Task Agents reports to a specific agentic application. The filter presents a dropdown of the available agentic applications, so you can isolate the performance of a single task agent rather than viewing all of them together.

To filter the Task Agents reports:

1. Click the Filters button.

2. Select the desired agentic application from the Agentic App Name dropdown, together with any other filters you need (Date range, Group, Channel, Language).

3. Click the View button to update the results.

4. Click Clear All to reset the filters.

Searching agentic applications in Monitor

The same Agentic App Name field is also available in OCP® Monitor, where a dedicated agentic search section lets you locate dialogs by agentic application. Select the agentic application from the Agentic App Name dropdown to retrieve the relevant dialogs for review. For more details on searching dialogs, see OCP® Monitor.

Voice Biometrics Reporting

Reporting in Voice Biometrics (VB) enables users to gather information on enrollments and verifications, including an Overall Enrollment Request graph, enrollment success rate, verification pass rate, and score distribution chart.

To access Voice Biometrics Reporting, navigate to the Voice Biometrics tab in Reporting.

Voice Biometrics Reporting Filters

Use the provided filters to refine your search as needed.

• Date range: Specify the date range for data extraction.

• Group: Select the group for data extraction.

• Verification origin: Choose either Agent (Live Agent) or IVR (Virtual Agent) .

• Region: Select a region if your environment supports multiple regions.

Enrollments report

• Overall requests: A graph showing successful and failed enrollment requests over the specified period.

• Success rate: Percentage of successful and failed enrollments.

• Cumulative results: Sum of successful, failed, and overall enrollments requests, over the selected date range.

Verifications report

• Overall requests: A graph displaying the number of passed and failed verifications over the selected date range.

• Success rate: The percentage of successful and failed enrollments. By default only TRUE HIGH are considered Passed verification, but users may change that to include TRUE_MEDIUM as well.

• Score distribution: The distribution of different verification results.

Liveness Detection Report

OCP Reporting now includes dedicated visibility into Liveness Detection outcomes at the call level. This helps teams understand how much suspicious voice traffic is reaching their applications, monitor trends over time, and filter results by operational dimensions such as group, region, and call origin.

Liveness Detection is a security capability designed to help determine whether the voice in a call appears to come from a live human speaker or from a spoofing attempt (such as synthetic or replayed audio).

It is intended to help detect multiple categories of spoofing, including:

• Text-to-speech (TTS) deepfake attacks

• Replay attacks, where recorded audio is played back

• Out-of-domain or contextually suspicious utterances used as an additional fraud signal

The new reporting is designed to provide clear access to:

• Call-level liveness outcomes (for example: genuine, spoofed, low-data, and related result categories)

• Counts and trends showing how many suspicious calls reached an application

• Filtering/segmentation by supported operational filters (for example group, region, call origin)

• Historical trend analysis across selected time periods

Blocklisting Report

OCP Reporting now includes dedicated visibility into Blocklisting outcomes at the call level. This helps customers understand how often known bad actors are being detected, whether activity is increasing over time, and where blocklisted calls are appearing across groups, regions, and supported origin dimensions.

Blocklisting identifies whether an incoming caller’s voice matches a voiceprint associated with a known fraudster or otherwise blocked identity. This helps organizations recognize repeat bad actors even if they attempt to call again using a different claimed identity

The reporting delivered in this scope is designed to provide:

• Call-level blocklist outcomes (match / no match)

• Visibility into counts and trends of blocklist matches over time

• Filtering/segmentation by supported operational dimensions (for example group, region, call origin)

• Trend-based reporting across selected periods to monitor recurrenceThis reporting update is about visibility and analytics. It does not change blocklisting decision logic during the call.

Agent Assist

Agent Assist reporting shows how live agents perform on calls that Agent Assist supports. It covers call volume, business flow outcomes, agent skill activity, and caller and agent sentiment for the selected group and period.

The Agent Assist reporting page is available in Insights → Reporting.

The page contains four sections:

• Call Volume: Overall call count and average handling time.

• Business Flows: Volume and outcome of business flow requests.

• Agent Skills: Skill coverage, agent participation, and per-agent detail.

• Sentiment Review: Call experience outcomes and sentiment polarity for callers and agents.

Accessing Agent Assist reporting

To open the Agent Assist page:

1. Navigate to Insights in the OCP® side menu.

2. Select Reporting.

3. Open the Agent Assist tab.

4. Select the group, date range, timezone, and any other filters required for your analysis.

5. Click the Apply button.

Agent Assist filters

Set the scope of every section on the page in the filters bar at the top.

The following controls are available:

• Group: The OCP® group that the metrics refer to. Select a single group, or several groups to report on them together.

• Date Range: The reported period. Choose a preset, such as Last week or Last month, or set a custom From and To date and time.

• Timezone: The timezone applied to the reported timestamps.

• Agent ID: The unique identifier of a live agent.

• Agent Skill: The agent skill group.

• Language: The language of the interaction, for example English US.

• Region: The OCP® environment origin region.

• Test Flag: Select true to fetch only test-flagged calls. Select false to exclude them. By default, all calls are reported.

To apply or reset the filters:

1. Click the Filters button and select the filters you want to pin.

2. Fill in the pinned filter fields as needed.

3. Click the Apply button to load the results.

4. Click the Clear all button to reset every filter field.

Each section also has its own time granularity control with the Month, Week, Days, and Hours options. The available options depend on the selected date range.

Cards labeled vs previous period compare the current value against the period of equal length immediately before the selected date range. A green value marks an improvement, and a red value marks a decline.

Call Volume

The Call Volume section reports the volume of Agent Assist calls and how long agents spend on them.

The following information is available:

• Overall Calls: The number of Agent Assist calls in the selected period, with the trend against the previous period.

• Avg Call Handling Time: The average call handling time in minutes, with the trend against the previous period.

• Overall Calls chart: The call volume over time, at the selected granularity.

Use this section to confirm the traffic volume behind the rest of the page before you interpret the other sections.

Business Flows

The Business Flows section reports the business flow requests that agents run during Agent Assist calls, broken down by outcome.

The following information is available:

• Volume: Three counters, each with the trend against the previous period.

  • Success: Business flows that completed.

  • Rejection: Business flow requests that ended in rejection.

  • Failure: Business flows that ended in failure.

• Business Flows chart: The Success, Rejection, and Failure volume over time. Click a legend checkbox to display a single outcome. Deselect all checkboxes to return to the initial state.

• Completion rate: The share of completed business flows over all business flows, shown as a percentage and as an x out of y count, with the trend against the previous period.

• Business Flow Requests: The number of requests per business flow intent, ranked by volume.

Use this section to identify the intents that agents request most and the flows that reject or fail most often.

Agent Skills

The Agent Skills section reports which skills handle Agent Assist traffic and how the agents behind those skills perform.

The following information is available:

• All Skills: The number of distinct agent skills in the selected scope.

• Agents Participation: The number of distinct agents in the selected scope.

• Top Skills: The 10 highest-volume skills. Select a skill to scope the section charts to that skill.

• Business flow volume over time: The business flow volume for the selected skill, at the selected granularity.

• Business Flow Volume: The average business flow volume per skill, with the trend against the previous period.

• Completion Rate: The share of completed business flows, with the trend against the previous period.

• Overall Calls: The reported call volume.

• Avg Call Duration: The average call duration in minutes.

• Agents Participate: The number of agents that participated in the reported calls.

Reviewing individual agents

Click the View full report button to open the Agent Skills full report. The full report keeps the section and adds a per-agent table.

The table contains the following columns:

• Agents ID: The identifier of the live agent.

• Calls: The number of calls the agent handled.

• Avg Duration: The average call duration in minutes.

• Resolution: The resolution rate of the agent.

• Skills: The number of skills assigned to the agent.

Click the arrow at the start of a row to expand it and list every skill of that agent under All Skills.

Sentiment Review

The Sentiment Review section reports call experience outcomes and sentiment polarity for the selected period and filters. It gives supervisors the polarity distribution, the Call Experience (XP) outcome counts, and the sentiment trend in one place.

Sentiment Review is available on the Agent Assist overview page only.

Call Experience KPI cards

Five cards report the number of calls per Call Experience (XP) outcome.

• Excellent XP: Calls with an excellent call experience. An increase counts as an improvement.

• Negative XP: Calls with a negative call experience. A decrease counts as an improvement.

• Recovery XP: Calls where the experience recovered during the call. An increase counts as an improvement.

• Poor XP: Calls with a poor call experience. A decrease counts as an improvement.

• Neutral: Calls with a neutral call experience. This card reports a count only, without a trend.

Caller sentiment

The Caller sentiment chart reports the share of Positive, Neutral, and Negative calls over time, at the selected granularity. Values are percentages of the caller sentiment recorded in each interval.

Click a legend checkbox to display a single polarity. Deselect all checkboxes to return to the initial state.

Sentiment breakdown

Two cards summarize the polarity distribution for the selected period:

• Overall caller sentiment: The polarity distribution of the caller side of the conversation.

• Agent responsiveness: The polarity distribution of the agent side of the conversation.

Each card reports the percentage and the call count for Positive, Neutral, and Negative, together with a stacked bar of the same three values.

How polarity maps to Call Experience outcomes

The breakdown cards and the Caller sentiment chart roll the Call Experience outcomes up into three polarities:

Best practices

• Check the Call Volume section first. Low call volume makes every other card on the page unreliable.

• Keep the date range narrow when you validate the effect of a recent deployment or configuration change.

• Read the Excellent XP and Recovery XP cards together with Negative XP and Poor XP. A rise in recovery alongside a fall in poor experience marks an improvement in agent handling.

• Compare Overall caller sentiment against Agent responsiveness to separate a difficult caller base from a slow agent response.

• Use the Agent Skill filter to confirm whether a sentiment change is limited to one skill group.

• Open representative dialogs in OCP® Monitor when a trend changes unexpectedly.

Agent Performance

Agent Performance reporting provides visibility into conversations handled by Agentic Flows. It helps you understand how each agent type performs, how often agentic steps are used, where conversations are routed, and which dialogs need further review in OCP® Monitor.

The Agent Performance page is available in Insights → Reporting for groups where Agentic Reporting is enabled.

Agent Performance metrics are computed from dialog data and are presented per Orchestrator agent type. This means that when a Concierge Agent delegates work to another agent, the delegated agent’s metrics are reported under that agent instead of being counted only under the Concierge Agent.

Accessing Agent Performance

To open the Agent Performance page:

1. Navigate to Insights in the OCP® side menu.

2. Select Reporting.

3. Open the Agent Performance tab.

4. Select the group, date range, and any other available filters required for your analysis.

5. Select the agent you want to review. The page automatically displays the reporting view that matches the selected agent type.

Agent Performance views

The Agent Performance page identifies the selected agent type and displays the relevant metrics for that role. Each view starts with a coverage summary, so you can see how much of the selected call volume includes agentic activity for that agent type.

• Q&A Agent: Shows coverage, headline KPIs, daily trend, and termination-by-outcome breakdown. Use it to measure knowledge-base question answering activity, completion, coverage, and contained self-service performance.

• Concierge Agent: Shows coverage, escalation queue-routing breakdown, and a PII-safe caller word cloud. Use it to understand how callers are being routed, which escalation queues are used, and what caller topics appear most frequently.

• Concierge-in-flow: Shows coverage, routing-destination frequency, selection KPI, daily routing timeline, and word cloud. Use it to review routing behavior, destination distribution, and routing trends over time.

• Task Agent: Shows coverage and daily invocation trend. Use it to track how often task agents are invoked and identify changes in usage over the selected period.

Q&A Agent reporting

The Q&A Agent view focuses on knowledge-retrieval interactions. It helps you evaluate whether the agent is being used, whether conversations are completed, and how often the Q&A interaction contributes to contained self-service.

The following information is available:

• Agent Coverage: The volume of dialogs containing Q&A agentic activity for the selected scope.

• Completion KPI: The completion performance for Q&A interactions.

• Contained self-service KPI: The percentage of conversations handled without transfer to a live agent, where applicable.

• Daily trend: A day-by-day view of Q&A activity over the selected period.

• Termination-by-outcome breakdown: A breakdown of how conversations ended, such as completed, terminated, transferred, or other available outcomes.

Concierge Agent reporting

The Concierge Agent view focuses on top-level routing and handoff behavior. A Concierge Agent typically greets the caller, determines the caller request, and delegates to a Task Agent, Q&A Agent, Concierge-in-flow, or escalation queue.

The following information is available:

• Agent Coverage: The volume of dialogs containing Concierge agentic activity for the selected scope.

• Escalation queue frequency - routing breakdown: The distribution of calls routed to configured escalation queues.

• Word cloud: A visual summary of frequent caller terms with personally identifiable information removed or protected.

Concierge-in-flow reporting

The Concierge-in-flow view focuses on routing-destination selection. It helps you understand where conversations are being sent and how routing activity changes over time.

The following information is available:

• Agent Coverage: The volume of dialogs containing Concierge-in-flow agentic activity for the selected scope.

• Routing-destination frequency: The number of times each destination was selected.

• Selection KPI: A summary metric for routing selection activity.

• Daily routing timeline: A trend view showing routing activity across the selected date range.

• Word cloud: A visual summary of common caller terms associated with routed interactions.

Task Agent reporting

The Task Agent view focuses on invocation volume for task-based agents. Task Agents execute a specific workflow, often using tools such as miniApps, web services, MCP servers, or announcements to complete their goal.

The following information is available:

• Agent Coverage: The volume of dialogs containing Task Agent activity for the selected scope.

• Daily invocation trend: The number of Task Agent invocations per day over the selected period.

Reviewing agentic dialogs in Monitor

Agentic Reporting works together with OCP® Monitor. When reviewing individual dialogs, Monitor can show agentic steps in the expanded dialog view.

For agentic steps, Monitor provides:

• Agent-type icons that make it easier to identify Q&A, Concierge, Concierge-in-flow, and Task activity in the dialog sequence.

• A Document ID deep link to Pathfinder for Q&A retrieval steps, where available.

• The retrieved file name, where available.

• An agentic type search filter for locating dialogs that include specific agentic activity.

• Inline Q&A judge scores for faithfulness and answer relevance, where available for the selected dialog.

Best practices

• Start with the coverage summary before interpreting KPIs or trends.

• Use narrow date ranges when validating the effect of a recent deployment or configuration change.

• Compare Concierge destination trends with your expected escalation and routing design.

• Use the Q&A termination breakdown to identify cases where knowledge retrieval may still lead to transfer or termination.

• Open representative dialogs in Monitor when a trend changes unexpectedly.

Current limitations

The first version of Agent Performance focuses on per-agent-type reporting and Monitor visibility. Some quality-evaluation cards and advanced drill-through capabilities are planned for future iterations and may not be visible in the current release.

Advanced Conversational Insights

More detailed advanced metrics for the full application history may be found in Advanced OCP Conversational Insights®.

Metrics API

Insights Report Metrics with corresponding filters are also available through the Metrics API.