Copilot Recommendations
Lists the recommendations shown to agents on the Copilot application and how agents interacted with them. For additional details regarding recommendations, please visit this article.
On the "Copilot Recommendations" report, you can find the following filters:
- "Agent Name".
- "Recommendations Interval date".
- "Recommendations Source".
- "Recommendations Title".
- "Recommendations Type".
The following fields are available in this report and for extraction:
Name | Description | Type |
---|---|---|
Agent Name | Name of the agent. | String |
Interval Date | Date when the recommendation was delivered to the agent. | Date (YYYY-MM-DD) |
Source | Source knowledge base of the recommendation. KM relates to Talkdesk Knowledge Management. | String |
Type | Type of recommendation. Possible values: - KM (Knowledge Management Card). - Quick answer (Knowledge Management Card triggered by intent). - NLQA (Answer produced by Question Answer algorithm). | String |
Intent | Intent that triggered the recommendation (applies to Quickanswer type). | String |
Title | Title of the article or card that was recommended. In the case of NLQA type, since the engine extracts a portion of the original article, the value in this column corresponds to the question identified by the recommendations engine. | String |
Delivered | Number of times that the recommendation was produced, for that agent, on that date. This value can be 0 if a given recommendation was produced but the agent did not see it because the Agent Assist application was closed. | Number |
Percentage of clicked | Percentage of recommendations that were clicked on/opened by the agent over delivered recommendations. | Number |
Percentage of negative feedback | Percentage of recommendations that received negative feedback over delivered recommendations. | Number |
Percentage of positive feedback | Percentage of recommendations that received positive feedback over delivered recommendations | Number |
Updated 4 months ago