Hi Andres, 

Thanks for writing in! I agree that ratings are key to evaluating customer service performance. In the conversation_history table, there should be an assignee_id alongside the rating information. The value for assignee_id for a given conversation should match the conversation-rating-object's "teammate" value at the rating creation time. Does this data work to satisfy your needs?

assignee_id
conversation_rating_created_at
conversation_rating_remark
conversation_rating_value

Entity Relationship Diagram


Do let us know if this helps!

Best,

Ray Harris

Product Manager, Sales & Customer Success Analytics

Hi Ray, thank you for replying so quickly!

Unfortunately, the assignee and the rated teammates can be different. The person being rated is the last person having sent a reply to a customer before the rating survey was sent, independently of who the conversation has been or is assigned to at the rating creation time (an example can be observed with conversation_id 105044200383875).

This is why having the teammate rated field would not only make rating attribution much easier to surface but it would also make the rating attribution fully reliable :)

I see, thanks for clarifying. So then if the rating gets assigned to the most recent teammate to send a message, then the current best available workaround must be to join conversation_parts to the conversation to find the author of the latest admin-comment up to the time of the rating creation. Yikes! 

We'll add it to our backlog to improve this experience. Thank you for the feedback!

Hi Ray, that is correct, with another nice subtlety: the author of the latest admin-comment up to the time of the rating creation may not be the rated agent (as displayed in Intercom's dashboards which are the reference for the teammates) since that time is the time when the customer rates the conversation - and there can be replies from other teammates in the meantime which won't be the rated teammate.

So I'm currently using a workaround (that's the one I came up with anyway) where I'm using the question asked by the bot (which vary by language) instead of the author of the latest admin-comment up to the time of the rating creation.

So thank you for adding this to your backlog, this will indeed improve the experience and reliability (since I won't have to remember to update my query if new languages are added)! :)