Workforce Engagement Management

Hey folks,

I hope that everyone is well today. 

I've recently started looking at Genesys Topics! Topics has me quite excited in terms of the insight / depth of analytical data the feature might make available to us. I decided to test the function on a 'busy' multi-channel queue (SMS, email and voice). I created:

• 10 topics relevant to the queue in question. Created around 600 phrases for each topic (med-low strictness), en-GB (dialect - UK-based) and 'Both' participants. Published.
• Created a Program, assigned the test queue in question. Assigned my topics. No AI insights. Published. 
• Created categories for each of my topics in order to 'tag' interactions accordingly. 

I've left this active for a few days and have today analysed the data. More than a little deflated. Perhaps I'm doing something wrong but this is what I have noted:

• For some reason, all emails within the test queue have been assessed using the default 'General' program for some reason, and not the program I created and assigned to the queue? Although, sentiment / topic analysis is working and visible within events within the interaction transcript - it's just utilising the incorrect program.
• Around 50% of the voice interactions within the queue, whist leveraging the correct program, have no 'Events' window displaying within the transcript and a sentiment rating of 0 / neutral - when there is clearly a lot sentiment clearly displayed within the transcript. It's almost as though the analysis hasn't taken place. 
• Whist leveraging the correct program, I haven't found one SMS interaction that is displaying an 'Events' window - they all show a sentiment rating of 0 / neutral.
• Of the interactions that have utilised the correct program and topics / sentiment analysis has taken place - Topics has made only 3 hits in four days. Having processed thousands of interactions. 2 of those hits were nonsense and had absolutely no association with any of the phrases within the flagged Topic. 

Am i doing something wildly incorrect? Had such high-hopes for this feature. Is anyone using it successfully? Any hints / tips / pointers?

Cheers.

#Reporting/Analytics

------------------------------
Matthew Tipler
------------------------------

Hi Matthew,

I am going to move this over to the Workforce Engagement Management Community so the Speech and Text experts can try and help resolve this

------------------------------
Sam Jillard
Online Community Manager/Moderator
Genesys - Employees
------------------------------

Original Message:
Sent: 11-24-2025 09:29
From: Matthew Tipler
Subject: Genesys Topics

Hey folks,

I hope that everyone is well today. 

I've recently started looking at Genesys Topics! Topics has me quite excited in terms of the insight / depth of analytical data the feature might make available to us. I decided to test the function on a 'busy' multi-channel queue (SMS, email and voice). I created:

• 10 topics relevant to the queue in question. Created around 600 phrases for each topic (med-low strictness), en-GB (dialect - UK-based) and 'Both' participants. Published.
• Created a Program, assigned the test queue in question. Assigned my topics. No AI insights. Published. 
• Created categories for each of my topics in order to 'tag' interactions accordingly. 

I've left this active for a few days and have today analysed the data. More than a little deflated. Perhaps I'm doing something wrong but this is what I have noted:

• For some reason, all emails within the test queue have been assessed using the default 'General' program for some reason, and not the program I created and assigned to the queue? Although, sentiment / topic analysis is working and visible within events within the interaction transcript - it's just utilising the incorrect program.
• Around 50% of the voice interactions within the queue, whist leveraging the correct program, have no 'Events' window displaying within the transcript and a sentiment rating of 0 / neutral - when there is clearly a lot sentiment clearly displayed within the transcript. It's almost as though the analysis hasn't taken place. 
• Whist leveraging the correct program, I haven't found one SMS interaction that is displaying an 'Events' window - they all show a sentiment rating of 0 / neutral.
• Of the interactions that have utilised the correct program and topics / sentiment analysis has taken place - Topics has made only 3 hits in four days. Having processed thousands of interactions. 2 of those hits were nonsense and had absolutely no association with any of the phrases within the flagged Topic. 

Am i doing something wildly incorrect? Had such high-hopes for this feature. Is anyone using it successfully? Any hints / tips / pointers?

Cheers.

#Reporting/Analytics

------------------------------
Matthew Tipler
------------------------------