The data model provided parses the transcript generated during the interaction, pulling out the following pieces of information:
- The text of the user queries and agent responses to those queries.
- Metadata associated with the session (e.g. intent, length, containment etc)
Armed with this information, you can begin to explore the answers to the following questions, which will help you tune agent interaction for maximum customer satisfaction:
- What types of questions are users asking about most frequently?
- To which types of questions does the bot reply most effectively to? Least effectively to?
- What types of questions require human intervention?
As with all Looker blocks, the solution is completely extensible, so you can incorporate any custom business logic or reports into the solution.
NOTE: This block requires exporting Dialogflow Log Files to BigQuery via Stackdriver.
Please review the step-by-step guide and ensure the following is complete before installing:
- Created a Dialogflow Agent
- Created a BigQuery Dataset to hold Dialogflow Stackdriver Logs and Looker Persistent Derived Table
- Created a StackDriver Sink Export to BigQuery