Workforce Engagement Management

Hi, can anyone help me!

In a QM evaluation form, is it possible for AI Scoring to detect an action within a specific time period during a voice interaction?

The scenario is as follows:

The client has an evaluation form with the following (Yes/No) question:

              Did the agent greet the customer within the first 5 seconds of the call? (AI Scoring should detect this; if the agent greets the customer after the 6th second, the answer should be No)

Thanks for support

Hi Ricardo,

From what I know, AI Scoring excels at analyzing the words spoken, not the timing of when they are spoken. For a greeting question, the official guidance is to phrase it to detect the presence of a standard greeting phrase, but I couldn't find anything in the documentation indicating that this is possible to detect an action within a specific time period.

Hi Camila!

I've tried doing this exercise-not using AI Scoring, but rather Categories with STA-but I haven't been able to create a rule that handles it accurately.

Here is an example:

Ricardo, 

I can't test it right now to confirm, but thinking about it, I SUPPOSE this might help you:

I guess that your rule cannot work because you're mixing two separate requirements (the topic and the time window) in a way that creates a logical conflict.

Can you try it and let me know if this worked?

I apologize in advance for suggesting something I haven't tested, but I can't have access to my lab environment at the moment.

This is an important point as some other customers are also looking at assessing and scoring questions based on handle time duration or silence duration and it doesn't seem to be supported. This question shouldn't be of type yes/no. Here's what the business users wanted iin my case: 

Question: How efficiently did the agent handle the call?
Score 5 if the call duration is below 10 minutes with any hold times clearly communicated and kept below 2 minutes. Score 4 if the call duration is below 10 minutes with hold usage communicated to the customer. Score 3 if the call duration is below 10 minutes and reasonable use of hold times. Score 2 if the call duration was between 10 and 20 minutes, repeated use of hold or pause, or hold times are in excess of 2 minutes. Score 1 if the call duration is above 20 minutes or excessive hold times causing frustration to the customer.

This is a pretty sophisticated question even if it may sound simple from a business perspective, it's not trivial to expect AI to just cover it simply by evaluating at transcript text.

You shouldn't assume that AI scoring would auto-detect this is Genesys doen't state this use case as supported. having implemented this, I know it is possible and has backend requirements if you do it with another solution available on Appfoundry (ping me if you want the details).

The transcript along with timers need to be evaluated so that the LLM would get both text and time differences between each turn and be able to assess the duration of handling or silence. This is a tricky and advanced use case but possible, however outside of Genesys current scoring from what I can see. So Ai scoring isn't always the same.