View utterance history

AI-generated summary

Genesys Cloud provides comprehensive bot conversation analysis through its Utterance History feature, accessible via the Optimization Dashboard in Architect with appropriate analytics and conversation viewing permissions. The platform enables users to view and analyze bot interactions with customers for up to ten days, supporting GDPR compliance through data retention limits. Users can filter interactions by specific action types—Ask for Intent, Ask for Yes/No, Ask for Slot, Wait for Input, Digital Menu, and Call Guide—as well as by outcome categories including Success, Confirmation prompted, and No Match. The system tracks multiple backend session outcomes across collection, confirmation, and disambiguation phases, including SUCCESS_COLLECTION, SUCCESS_CONFIRMATION_YES/NO, NO_MATCH_COLLECTION, NO_MATCH_CONFIRMATION, agent requests, confirmation prompts, system errors, input timeouts, expression errors, and disambiguation states. Voice Survey flows include SKIPPED_COLLECTION outcomes, while virtual agent functionality supports PARTIAL_COLLECTION for AI-powered slots collected incrementally across conversation turns. The platform captures Natural Language Understanding (NLU) confidence levels, validation failures, and expression evaluation errors. Users can click individual utterances to view complete conversations and access full Interaction pages in new browser tabs, with the ability to navigate between core bot flows and nested subflows. Complete audio conversations and transcripts are accessible from the Interaction Details tab, supporting troubleshooting and conversation context review. Audio replay functionality includes play, pause, volume adjustment, and playback speed controls (0.5x to 2x), with waveform visualization for navigating to specific recording segments. For agentic virtual agent flows, the platform enables detailed session analysis with session IDs, timestamped utterances, action calls, tool execution metrics, latency data, knowledge tool answers with source citations, and guardrail violation tracking. Users can access replay mode for full session visualization and open agents in AI Studio for further analysis. Role-based access controls require permissions including Analytics > botFlowReporting Turn > View and Analytics > ConversationDetail > View for accessing complete conversation details. The system supports multiple disambiguation and collection states within virtual agent flows, and provides feedback mechanisms for users to report article accuracy and helpfulness across role-based categories including agents, administrators, developers, partners, supervisors, and support personnel.

PrerequisitesClick to expand

In Architect, you can view the bot flow’s utterance history and complete details about a conversation. The Utterance History page returns a list of utterances, or sample phrases, that the bot receives from customers. Utterance history is available for ten days. Each entry includes these metrics:

The utterance that the bot received. For example, a numeric postal code or a “Check account balance” query.
The date that the bot received the utterance.
The bot’s session ID.
The action and action ID that the bot processed. For example, “16 Digital Menu” or “13 Ask for Slot.”
An action filter that enables you to sort by all actions or a specific action.
The bot outcome. For example, “Success,” “Confirmation prompted,” or “No Match.”
An outcome filter that enables you to select which outcomes by which to filter the list.
Bot details, such as the matched intent, where applicable.

View utterance history for the current bot flowClick to expand

Open the bot flow that contains the utterance history to view.
Under Insights and Optimization, click Optimization Dashboard.
Click Utterance History . The view populates with utterance details.
To view the complete conversation that contains the utterance, click the utterance.
Note: When viewing the utterance history of a core bot flow in a nested bot flow hierarchy, you can click the subflow name to open the subflow. Similarly, when viewing the utterance history of a subflow, you can click the core flow name to open the core flow.
To close the conversation and return to the utterance history, click Close .

View utterance history for specific actions in the bot flowClick to expand

You can view the utterance history for the following actions:

To view history from any of these actions in the flow, follow these steps:

Open the appropriate bot flow.
In a configured state or task, click an action that includes the utterance history you want to view.
Click More .
From the menu that appears, click Utterance History.
To view the complete conversation that contains the utterance, click the utterance.

View utterance history for an agentic virtual agent flowClick to expand

In Architect, you can view the utterance history of a bot or digital bot flow that is enabled with Virtual Agent and includes a Call Agentic Virtual Agent action. For more information about agentic virtual agents, see About agentic virtual agents and Connect agentic virtual agents to Architect Virtual Agent flows.

Note: To support GDPR compliance, utterance and flow execution data generated by the agentic virtual agent within Architect is retained for a maximum of 10 days.

You can review the following details:

The session ID unique to an agentic virtual agent interaction.
All utterances, along with their outcomes, action calls, and time stamps.
The complete conversation containing the utterance, along with turn-level metadata. You can drill down into each turn to review:
- The sequence of tool calls made during an interaction.
- Both data action and knowledge tool calls.
- The latency and execution status of each tool call.
- Answers generated by each knowledge tool call with source citations.
- Any guardrail violations that occurred during the interaction.
- All actions triggered by the agent.

View more details

To view the entire replay of the session, click More on the session ID and select Open in Replay Mode. For more information, see Use replay mode to troubleshoot an Architect flow.

To open an agent in AI Studio, click More on the agent name and select Open in AI Studio.

View complete details about an audio conversationClick to expand

You can view a complete audio conversation, including bot responses, from an interaction. Doing so can monitor the bot performance and help you troubleshoot or understand conversation details. For example, a full conversation can provide context for any transcription or audio issues.

View the full audio conversation from the utterance history view or interaction’s Details tab in Genesys Cloud. The utterance history view shows the recording details such as start and end time stamp, total recording length, and the current time position.

Open the bot flow that contains the utterance history to view.
Under Insights and Optimization, click Optimization Dashboard.
Click Utterance History . The view populates with utterance details.

To view the complete conversation that contains the utterance, perform the following steps:

Click the utterance. If a recording of the conversation is available, the audio recording appears as a waveform.

To listen to the audio, use the following replay controls:

Replay control	Description
Play	Starts or resumes playback.
Pause	Temporarily stops playback while keeping the current position.
Volume	Adjusts playback volume.
Playback speed slider	Plays the recording slower or faster (for example, 0.5x, 1x, 1.5x, or 2x).

To listen to a specific part of the recording, click the corresponding point in the audio waveform.

To access the entire conversation, click View Interaction . The Interaction page for the conversation opens in a new browser tab. For more information, see View an interaction’s Details tab.
To close the conversation and return to the utterance history, click Close .

Bot session view outcomesClick to expand

Backend outcome	Displays as	Description
SUCCESS_COLLECTION	Success	Input utterance matches the intent and slot with high confidence and does not trigger confirmation.
SUCCESS_CONFIRMATION_YES	Confirmation Yes	The user inputs “yes” during confirmation.
SUCCESS_CONFIRMATION_NO	Confirmation No	The user inputs “no” during confirmation.
NO_MATCH_COLLECTION	No Match	None response from NLU when expecting intent and slot. An intent response with low confidence. Triggers a validation failure.
NO_MATCH_CONFIRMATION	No Match Confirmation	Expecting yes or no during the confirmation phase and the user inputs something other than those answers.
AGENT_REQUESTED_BY_USER	Agent Request	The user inputs “I want to speak to an agent” Note: Not applicable to Voice Survey flows
CONFIRMATION_REQUIRED	Confirmation Prompted	The user inputs an utterance that triggers the confirmation phase, indicating confirmation required now
ERROR	System Error	Expression evaluation fails / NLU unexpected 500 error, and so on.
NO_INPUT_COLLECTION	No Input	For example: A user didn’t speak any response during the intent / slot collection phase
NO_INPUT_CONFIRMATION	No Input Confirmation	For example: A user didn’t speak any response during the confirmation phase
EXPRESSION_ERROR	Expression Error	Expression error encountered (for example, 1 divided by 0)
DISAMBIGUATION_REQUIRED	Disambiguation Required	Note: Not applicable to Voice Survey flows
SUCCESS_DISAMBIGUATION	Success Disambiguation	Note: Not applicable to Voice Survey flows
SUCCESS_DISAMBIGUATION_NONE	Success Disambiguation None	Note: Not applicable to Voice Survey flows
NO_MATCH_DISAMBIGUATION	No Match Disambiguation	Note: Not applicable to Voice Survey flows
NO_INPUT_DISAMBIGUATION	No Input Disambiguation	Note: Not applicable to Voice Survey flows
SKIPPED_COLLECTION	Skip	Note: Only applicable to Voice Survey flows
PARTIAL_COLLECTION	Partial Match	Virtual Agent. For AI-powered slots, a slot value can be partially collected turn by turn.

[NEXT] Was this article helpful?

Get user feedback about articles.

Loading...

View utterance history

[NEXT] Was this article helpful?