I want to determine Customer Sentiment statistics by day and interval for individual and groups of Agents.

To do this, I added Sentiment Instances, Positive Instances, and Negative Instances counts to the Agent Performance View. To see these stats for one agent for a particular day, I filter for one Agent and set the time to a single day (e.g., August 1, 2024). The summary row at the top totals all of the sentiment items for the day and the intervals below show the sentiment items by interval. 

This would be great except that when I manually go into each interaction's transcript and add up the sentiment counts that should be associated with the agent, my manual count conducted through the analysis of the transcripts does not match the daily total in the Agent Performance View for that agent. 

Some of the calls involve multiple agents, on-hold, and conferencing events. So, for those transcripts, I am reviewing them closely to only count sentiment items that are associated with the agent and not other agents, conferences, etc. But no matter what I try, I cannot get my manual counts to match. In most cases, the manual count of positives exceeds the count shown in the Agent Performance View and the manual count of negatives is less than what is in the Agent Performance View.

I have worked with 2 other analysts who conducted their manual counts for other agents on different full days and in all cases the manual counts do not match the counts displayed on the Agent Performance View. We cannot get our manual counts to match the Performance View values and do not see a pattern.

So, I guess my ultimate question is how are the Sentiment Instances tallied on the Agent Performance View report? The Resource Center states that sentiment instances are tracked and counted in the interval when the sentiment is recognized. But that doesn't seem to be the case.

#SpeechandTextAnalytics
#Reports

------------------------------
Dan Fontaine
C1
------------------------------

Hi Dan,

Can you share some screenshots with us? You can blur out any sensitive data.

How many positive/negative counts are you getting in Agent Performance vs Transcripts?

Also, are you able to create a CARE ticket so the Genesys team can investigate the data? You can add a note in your ticket the Product Manager Anik Dey requested a ServOps to investigate this.

I want to determine Customer Sentiment statistics by day and interval for individual and groups of Agents.

To do this, I added Sentiment Instances, Positive Instances, and Negative Instances counts to the Agent Performance View. To see these stats for one agent for a particular day, I filter for one Agent and set the time to a single day (e.g., August 1, 2024). The summary row at the top totals all of the sentiment items for the day and the intervals below show the sentiment items by interval. 

This would be great except that when I manually go into each interaction's transcript and add up the sentiment counts that should be associated with the agent, my manual count conducted through the analysis of the transcripts does not match the daily total in the Agent Performance View for that agent. 

Some of the calls involve multiple agents, on-hold, and conferencing events. So, for those transcripts, I am reviewing them closely to only count sentiment items that are associated with the agent and not other agents, conferences, etc. But no matter what I try, I cannot get my manual counts to match. In most cases, the manual count of positives exceeds the count shown in the Agent Performance View and the manual count of negatives is less than what is in the Agent Performance View.

I have worked with 2 other analysts who conducted their manual counts for other agents on different full days and in all cases the manual counts do not match the counts displayed on the Agent Performance View. We cannot get our manual counts to match the Performance View values and do not see a pattern.

So, I guess my ultimate question is how are the Sentiment Instances tallied on the Agent Performance View report? The Resource Center states that sentiment instances are tracked and counted in the interval when the sentiment is recognized. But that doesn't seem to be the case.

#SpeechandTextAnalytics
#Reports

------------------------------
Dan Fontaine
C1
------------------------------

I did open a ticket with Support. They are looking into it.

Agent Detail Shows 26 Conversations and the sentiment counts:

Support provided data from the API that is used in the Performance View. However, their API result only had 23 conversations for the day so I have asked them about that. Because there should be 26.

Going into each of the 26 conversations and counting the sentiment items associated with the agent results in a different total for the day shown in the Agent Detail report (above). The analysis is shown below. The 3 conversations in blue are the ones that were not returned in the API request that support ran.

Thank you.

Hi Dan,

Can you share some screenshots with us? You can blur out any sensitive data.

How many positive/negative counts are you getting in Agent Performance vs Transcripts?

Also, are you able to create a CARE ticket so the Genesys team can investigate the data? You can add a note in your ticket the Product Manager Anik Dey requested a ServOps to investigate this.

I want to determine Customer Sentiment statistics by day and interval for individual and groups of Agents.

To do this, I added Sentiment Instances, Positive Instances, and Negative Instances counts to the Agent Performance View. To see these stats for one agent for a particular day, I filter for one Agent and set the time to a single day (e.g., August 1, 2024). The summary row at the top totals all of the sentiment items for the day and the intervals below show the sentiment items by interval. 

This would be great except that when I manually go into each interaction's transcript and add up the sentiment counts that should be associated with the agent, my manual count conducted through the analysis of the transcripts does not match the daily total in the Agent Performance View for that agent. 

Some of the calls involve multiple agents, on-hold, and conferencing events. So, for those transcripts, I am reviewing them closely to only count sentiment items that are associated with the agent and not other agents, conferences, etc. But no matter what I try, I cannot get my manual counts to match. In most cases, the manual count of positives exceeds the count shown in the Agent Performance View and the manual count of negatives is less than what is in the Agent Performance View.

I have worked with 2 other analysts who conducted their manual counts for other agents on different full days and in all cases the manual counts do not match the counts displayed on the Agent Performance View. We cannot get our manual counts to match the Performance View values and do not see a pattern.

So, I guess my ultimate question is how are the Sentiment Instances tallied on the Agent Performance View report? The Resource Center states that sentiment instances are tracked and counted in the interval when the sentiment is recognized. But that doesn't seem to be the case.

#SpeechandTextAnalytics
#Reports

------------------------------
Dan Fontaine
C1
------------------------------