Good afternoon, we are creating a POC for a potential customer that wants to see their tech support cases imported into Customer Service (CE) and have Customer Insights do it's AI thing and group them together into Topics then name the topics?
The issue is that the cases that we imported from the customer have a case # and a description. It looks like whomever did the import (and or provided the data from the client) imported the customer data and now the Case NAME and ID are the same number. From what I can tell from the documentation, it looks like the Topic NAME is derived from some AI driven algorithm that looks at all the case TITLES that are grouped together??? All of our topics names have a number 4332211 that is from one of the cases that were grouped. It looks like it is just picking a random number from the group for the title, is that a fair assumption? To fix this, I'm guessing we will need to have a relevant CASE TITLE to import along with the CASE ID and description?
On the topic of how it groups the cases into topics, does anyone know what data it looks at to do this? I'm guessing the case DESCRIPTION?
One last question, if we update all the case names with actual relevant names, I'm guessing that the AI will see the change and update the topics?
Sorry for all the newbie questions -- trial by fire is the way this POC is getting built!
Thanks, Ford Wilkinson
Rushil, thanks for the reply and the information! On the topic of using the analytics experience in the "Core Customer Service apps", you are referring to the module embedded in CE?
Hi Fred – thanks for the questions!
The purpose of topics is to group similar cases together when they are semantically similar so that users can understand how different groups of cases are impacting key KPIs.
The data field that is used for topic discovery is configurable. We recommend using a field where you are most likely to find content related to the issue that the customer reached out to support with. For Case topics, this can be any string field on the case table. If you are to update the case titles with relevant data, the topics will update to reflect that.
One other note taking a look at your screenshot – it looks like you’re using the Customer Service Insights standalone application. The standalone application is on path for deprecation by December 8th, 2021 (notice here: Deprecations in Dynamics 365 Customer Service | Microsoft Docs). I would recommend using the similar historical analytics experience in the core Customer Service apps. You can get started here: Customer Service Insights | Microsoft Docs
For any additional questions relating to Insights in Customer Service, feel free to use the following forum: Dynamics 365 Customer Service - Forums, Blogs, Support
Cheers,
Rushil
André Arnaud de Cal...
291,979
Super User 2025 Season 1
Martin Dráb
230,848
Most Valuable Professional
nmaenpaa
101,156