The Solve Chat Analytics feature provides Forethought administrators with access to detailed reports and insights at both the chat and intent levels. This supplements the solve chat analytics tab and workflow-level analytics dashboard that we already have. This analytics view enables admins to:
- Get an overview of chat metrics including CSAT
- Visualize chat data
- Intent metrics
- Download data connected to chat conversations including CSAT data
Note: All analytics, including Looker tables and insights tables, are processed every 1 to 2 hours.
Analytics dashboard definitions
| Metric Name | Definition |
| Total chats | Number of chat conversations that were initiated in the selected time period. |
| Total chat deflections | Number of deflected conversations in the selected time period |
| Chat deflection rate | (Number of deflected conversations / Total number of conversations) x 100 (in the selected time period) |
| Chats that included one or more article suggestions | Percentage of chat conversations where a knowledge article was shown to the user. |
| Average Chat CSAT Score |
Average CSAT score |
| Visualization | What does it show? |
| Median chat duration |
The median duration a user spent in the chat. X axis shows the date. Y axis shows the time in seconds. |
| Number of chat conversations and number of chat conversations successfully deflected |
This shows a graph where X axis shows the conversation start hour (Ex: 2023-08-09 14) Y axis shows the number of chat conversations as well as number of chat conversations deflected. |
|
Chat CSAT score distribution (Bar chart) |
This shows the CSAT distribution bar chart. X axis shows the number of conversations for which CSAT score was given by a user Y axis shows the CSAT score (Scale 1 to 5) that the user gave at the end of the chat conversation in the selected time period. |
|
Last predicted intent (Pie chart) |
This shows the top 10 intents in a pie chart. 9 of these are actual top 9 intents and the rest of the intents are clubbed into the other bucket. |
Data tables
There are 4 data tables which shows you raw data
- Chat CSAT data
- Chat intent metrics
- Knowledge article surfaced in chats
- Chat details
Chat CSAT data
This table shows the raw CSAT feedback data with the chat transcript. This only has conversations that have CSAT data. If CSAT is not enabled, this table will remain blank.
| Column name | Definition |
| Chat ID | This is the unique chat identifier (Forethought generated) You can use this to join with the Chat details table. |
| Chat start time |
Time when the chat was started Format: YYYY-MM-DD hh:mm:ss |
| Last predicted intent | This refers to the last predicted intent in the chat conversation. |
| Deflected |
Shows if the chat conversation was deflected or not. 1 - Chat was deflected 0 - Chat was not deflected |
| CSAT score |
CSAT score the user submitted. (Scale - 1 to 5) |
| CSAT Feedback | Additional CSAT feedback provided by the user for this conversation |
| Chat duration |
The duration a user spent in this chat conversation. (seconds) |
| Chat transcript | The complete chat conversation transcript |
| Context variables used | All the context variables used in this chat conversation |
| Clicked on talk to an agent |
Shows if the user clicked to talk to an agent 1 - Clicked on talk to an agent 0 - Didn’t click on talk to an agent |
| Chat handoff |
Shows if chat handoff was successful. 1 - Successful chat handoff 0 - Unsuccessful chat handoff null - no chat handoff data |
Chat intent metrics
| Column name | Definition |
| Last predicted intent | Intent name |
| Number of chats | Number of times when this particular intent was the last predicted intent in a chat. |
| Intent deflection rate | (Number of conversations deflected in this intent / Total number of conversations predicted in this intent) x 100 |
Knowledge article surfaced in chats
| Column name | Definition |
| Title | Article title/heading |
| Source link | Link to the article |
| Number of times surfaced | Number of times this article was shown in the chat to the users in the selected time period |
Chat details
| Column name | Definition |
| Chat ID | This is the unique chat identifier (Forethought generated) You can use this to join with the Chat CSAT data |
| Chat start time |
Time when the chat was started Format: YYYY-MM-DD hh:mm:ss |
| Last predicted intent (user input) |
This shows the last user input in the chat conversation. Ex: Entered-via button or Go-To intent action |
| Last predicted intent | This refers to the last predicted intent in the chat conversation. |
| User query for last predicted intent | This shows the user query on which the last intent prediction was made. |
| Deflected |
Shows if the chat conversation was deflected or not. 1 - Chat was deflected 0 - Chat was not deflected |
| Generative AI used |
Shows if generative AI was used in the chat responses to user queries 1 - Yes generative AI was used 0 - No generative AI was not used |
| Chat duration |
The duration a user spent in this chat conversation. (seconds) |
| Chat transcript | The complete chat conversation transcript |
| Context variables used | All the context variables used in this chat conversation |
| Clicked on talk to agent |
Shows if the user clicked to talk to an agent 1 - Clicked on talk to an agent 0 - Didn’t click on talk to an agent |
| Chat handoff |
Shows if chat handoff was successful. 1 - Successful chat handoff 0 - Unsuccessful chat handoff null - no chat handoff data |
| Helpdesk ticket ID |
If at the end of a conversation, the user wanted to create a ticket, this shows the ticket id created in your helpdesk. null signifies no ticket was created. |