Based on your question, I guess you should use Logs/
examples on the API Reference for base what you CSV file gonna have, for example: inputText, outputText, conversation_id, dateConversation, intents, entities, context, workspace_id, etc.
In this example using Python that will return these data (you can also use another language that has HTTP request support):
List logs:
list_logs(self, workspace_id, sort=None, filter=None, page_limit=None, cursor=None)
Request:
import json
import watson_developer_cloud
conversation = watson_developer_cloud.ConversationV1(
username='{username}',
password='{password}',
version='2018-02-16'
)
response = conversation.list_logs(
workspace_id = '9978a49e-ea89-4493-b33d-82298d3db20d'
)
print(json.dumps(response, indent=2))
Obs.: Take a look at this example saving inside one txt
file the conversation Logs, you can also check this answer to use the same logic to save in one csv
file.
Obs.: Nowadays Watson conversation changed the name for Watson Assistant.