Conversations dataset
This dataset provides details at the conversation level, including the conversation ID, its current state, the communication channel used, and the assignee. It focuses on the overarching aspects of a conversation but does not cover individual conversation actions.
Metrics
Metric Name | Description |
New conversations | The number of all new inbound and outbound conversations.
It only includes conversations where the first customers message was received during the selected date range. |
AI Answer resolution rate | The percentage of conversations where Fin AI Agent's last answer was an AI Answer, and the customer either replied with positive feedback or didn’t ask to talk to a teammate (out of all Fin involved conversations).
|
AI Answer resolutions | The conversations where Fin AI Agent's last answer was an AI Answer, and the customer either replied with positive feedback or didn't ask to talk to a teammate.
|
Fin AI Agent pending conversations | The number of conversations where Fin was involved and the conversation is still in progress (not resolved by Fin, not escalated to a teammate, and not in procedure handoff).
A conversation is pending when either:
Note: Pending is a current state. Conversations can move out of Pending if the customer replies later, or if Fin escalates or resolves the conversation.
|
Fin AI Agent pending rate | The percentage of Fin-involved conversations that are still in progress (not resolved by Fin, not escalated to a teammate, and not in procedure handoff) out of all the Fin AI Agent involved conversations.
A conversation is Pending when either:
Note: Pending is a current state. Conversations can move out of Pending if the customer replies later, or if Fin escalates or resolves the conversation.
|
Fin AI Agent answer rate | The percentage of conversations in which Fin AI Agent sends at least one answer, whether it's an AI Answer or a Custom Answer, out of all Fin AI Agent involved conversations. An answer from Fin AI Agent is defined as any content that could potentially answer the customer's question. |
Fin AI Agent answered conversations | The number of conversations in which Fin AI Agent sends at least one answer, whether it's an AI Answer or a Custom Answer. An answer from Fin AI Agent is defined as any content that could potentially answer the customer's question. |
Fin AI Agent assumed resolutions | The number of conversations in which the customer left without asking to talk to a teammate after the last answer was delivered by Fin AI Agent. |
Fin AI Agent confirmed resolutions | The number of conversations in which the customer replied with positive feedback after the last answer was delivered by Fin AI Agent. |
Fin AI Agent deflected conversations | The number of Fin AI Agent involved conversations where either:
|
Fin AI Agent deflection rate | The percentage of conversations in which Fin AI Agent got involved and either:
out of all Fin AI Agent involved conversations.
|
Fin AI Agent involved conversations | The number of new conversations that were passed to and handled by Fin AI Agent, where Fin actively engaged even briefly (for example, by providing a headline response or escalating the conversation). Note:
|
Fin AI Agent involvement rate | The percentage of new conversations that were passed to and handled by Fin AI Agent, where Fin actively engaged — even briefly (for example, by providing a headline response or escalating the conversation) — out of all new conversations. Note:
|
Fin AI Agent resolution rate | The percentage of conversations resolved by Fin AI Agent, divided by conversations where Fin AI Agent was involved.
A resolved conversation constitutes either a confirmed OR assumed resolution.
A confirmed resolution is when a customer:
An assumed resolution is when a customer:
|
Fin AI Agent resolved conversations | The number of conversations where Fin AI Agent provided at least one response (AI Answer or Custom Answer), and following the last response: The customer indicates satisfaction by:
OR The customer exits the conversation without indicating further need for assistance, as shown by:
|
Fin AI Agent escalated conversations | The number of conversations where Fin was involved and the conversation was escalated to a teammate (either because the customer asked to speak to a person, or because your configuration escalated the conversation). |
Fin AI Agent escalated rate | Percentage of conversations in which Fin was involved and the conversation was escalated to a teammate (either because the customer asked to speak to a person, or because your configuration escalated the conversation) out of all Fin Al Agent involved conversations. |
Fin AI Agent configuration based escalation rate | Percentage of escalated conversations that were triggered by configuration settings (for example, escalation rules/guidance) out of all Fin Al Agent escalated conversations. |
Fin AI Agent unanswered conversations | The number of Fin AI Agent involved conversations where either:
|
Fin AI Agent unresolved conversations | Number of conversations in which Fin Al Agent got involved but customer either gave negative feedback or asked to talk to a teammate. |
Procedure handoff | The number of conversations in which the Fin Procedure or Fin Task intentionally handed the conversation to a team or another workflow. These are configured handovers within your procedure or task (for example, a “Hand off to Billing Team” action or an instruction within your Procedure Guidance settings). |
Custom Answer resolutions | Conversations where Fin AI Agents last answer was a Custom Answer and the customer either replied with positive feedback or didn't ask to talk to a teammate. |
Custom Answer resolution rate | Percentage of conversations where Fin AI Agents last answer was a Custom Answer and the customer either replied with positive feedback or didn’t ask to talk to a teammate (out of all Fin AI Agent involved conversations). |
Attributes
Fin AI Agent: pending reason
| Use this to understand why a conversation is currently Pending.
|
Fin AI Agent: configuration based escalation reason | Use this to isolate escalations that happen due to configuration (rather than a customer explicitly requesting a teammate):
|
Conversation rating dataset
This dataset provides details at the conversation rating level, including the conversation rating id, the conversation rating, and the conversation rating remark.
Metrics
Metric Name | Description |
Fin AI Agent conversation ratings | Number of conversation ratings for Fin AI Agent. |
Fin AI Agent CSAT score | Percentage of positively rated (🤩 or 😀 ratings) Fin AI Agent conversations out of all Fin AI Agent conversations rated. |
Fin AI Agent DSAT score | Percentage of negative (😠, 🙁 or 😐) conversations ratings for Fin AI Agent out of all conversations ratings for Fin AI Agent. |
Fin AI Agent negative conversation ratings | Number of negative (😠, 🙁 or 😐) conversation ratings for Fin AI Agent. |
Fin AI Agent positive conversation ratings | Number of positive (😀 or 🤩) conversation ratings for Fin AI Agent. |
Copilot dataset
This dataset captures teammate interactions with Copilot on your workspace. It allows you to understand how Copilot is being used by teammates during conversations.
Metrics
Metric Name | Description |
Copilot questions | Number of questions asked to Copilot. |
Teammates using Copilot | Number of teammates that have used Copilot at least once. |
Copilot copied answer rate | Percentage of conversations in which Copilot provided an answer that was copied or added to the composer. Only conversations with at least one teammate reply and one customer reply are included. |
