CUE Trends Widgets

Call Volume

Call Volume plots total number of calls for a specific time period. This widget also provides the option to select a custom date range.  

Depending on the date range chosen the units of measurement will change.

Time periods of one day, including comparisons with “Today” - the reference line is in hours of the day.

For all time periods beyond one day - the reference line changes to dates (month/day).

This widget updates at the completion of an hour.

Recognitions

This widget displays the number of times callers traversed each dialog state (Visits) and the percentage of responses that were In Grammar, No Match and No Inputs. Click on a percentage number (%) to view raw counts. 

If there is a number greater than zero in the "In Grammar" column for a given Dialog State name, click on a dialog state name to drill down into the Interpretations level.  This is where CUE shows what the recognitions engine understood the response to be. Click in the gray box with the dialog state name to return to the initial view. This widget will update after a call has ended. 

3 instances when Interpretations are not available

1. The developer has blocked the details from the interpretation because it may be sensitive. Many times you’ll see XXXX or ? in the interpretation level because the details of the interpretation is intentionally being redacted from posting to the database. We do this for things like SSN and credit card numbers.

2. The developer has not coded the app to send interpretation details to CUE. This is to be expected in some cases as this is a new feature. We’re reviewing all applications and in many cases, we’ve already started to work with your account teams to schedule in an update to correct this problem. In the instance, you’ll see a “?” or a message saying there are no interpretations for this dialog state.

3. The grammar is too large, meaning it has more than 200 possible unique interpretations, we will stop aggregating individual responses at the 201st unique interpretation. An example of this is a grammar for Payment Amount. In cases where the caller may enter in any amount, this can get out of hand. While we are still saving each individual interpretation in the database, for the purpose of aggregation and tracking customer behavior, we weren’t sure how beneficial it would be to know that one caller said $31.84 and another said “$31.85”

Instead, we still provide the percentage breakdown between In Grammar (They said $31.84 and we recognized it), No Match (They said “thirty-oneFido stop chasing the cat!” and we weren’t sure what to interpret.) and No Input (Silence detected on the line)

Call Outcomes

Outcomes are defined in the application and aggregate in this widget upon the completion of a call. Sometimes it's necessary to compare the same outcome over two different time periods at the same time, which is why we offer the option to duplicate or add a second Call Outcomes widget.

There are two views, Total % and Over Time. The Over Time view allows for up to three outcomes to be graphed. Click on the plus to add another outcome. Click on an outcome name to remove it. The background color of the outcome name determines the plot color on the graph.  Call Outcomes with no activity for a time period will not appear in the graph. 

The Total % View is designed to summarize and tabulated call outcomes for a given time period.

The benefits of being able to duplicate the Call Outcomes widget:
1. will allow you to simultaneously view two different time periods on the same screen.
2. will let you compare the percentage difference of each outcome between time periods. When in Duplication Mode, the Call Outcomes widgets will have an additional column for comparing percentage difference between the two time periods.
3. will allow you to view the impact of app changes affecting outcomes in fewer steps. 
 
A quick screen shot and you're ready to insert the results into an email and share with the team. 
 

 

Hang ups

Hang Ups summarize the total number of hangups in the application for the time period selected, what percentage of the total calls were hangups and how many of the hang ups were "short calls" as determined by the threshold set by the user (gear icon). 

Below the time period selection displays the dialog state callers were in when the call ended via hangup.

No Match will increment when the caller speaks something that the recognition engine does not have in grammar.

No Input will increment when the caller speech is not detected.