Quality Analysis: Categories
- Product Knowledge
On this page |
AVAILABLE FROM VERSION 3.36.0
What’s about
Categories are groups of filters, useful to analyse voice recordings only on Post Call Analytics
By default all created categories are automatically applied to Call Analytics transcripts. Remember that to apply categories to your transcripts, you must create them before submitting the transcript request, as categories cannot be applied retroactively.
Call Analytics categories are composed of rules. For each category, you must create between 1 and 20 rules.
Requirements
Feature available from version 3.36.0 and developed on New Client Experience interface
Categories are available only for AWS
To create categories, you need to enable on your license Quality Analysis
Configuration
From QA Categories section, you can view all created categories and the relative provider.
The unique key consists of provider + unique name
First of all you have to select the AWS Region
And then you can create a new category, by clicking on Add category
And insert:
Unique name (it is considered as a key, so spaces are not accepted, you can only insert underscore)
AWS Region (by default the system will insert the one selected before, but it’s possible to change it)
AWS Provider
Rules to define that can be of 4 types (described below)
Types of rules
You can configure 4 types of rules
If you create more rules, they must all occur in order to track the desired elements in the voice recording
Interruption
This rule marks the presence or absence of interruptions in Call Analytics transcript output. It is useful to check whether the agent behaves correctly or whether the client is aggressive during the conversation, by interrupting the agent
These rules are designed to correspond to:
Cases where an agent interrupts a customer
Cases in which a customer interrupts an agent
One of the participants interrupts the other (by choosing both - default option)
Lack of interruptions (by selecting Negate option)
Applicable Fields | Description |
|---|---|
Negate | Enable it to mark voice recordings that do not contain interruptions. Leave it disable to mark voice recordings that contain interruptions. |
Time Range | Absolute = Allows you to specify a time interval (in milliseconds) in the audio, during which you want to search for an interruption Relative = Allows you to specify a time interval (in percent) in the media file, during which you want to search for an interruption |
Role
| Specify if you want to search interruptions Agent side, Customer side or Both. By default the system is set on Both participants |
Threshold | Specify the duration of the interruptions in milliseconds. For example, you can mark a speech that contains more than 10.000 milliseconds of interrupts. |
Interval
| You can explore paragraph below |
Non-Talk Time
This rule represents the presence or absence of silent periods in the Call Analytics transcript output, by respecting the set mandatory threshold. During a conversation, physiological pauses are normal, but by setting the rule, the system tracks if there are longer pauses (e.g., if the customer has been put on hold)
These rules are designed to match:
The presence of silence at certain times during the call
The presence of voice at certain times during the call (by selecting Negate option)
Applicable Fields | Description |
|---|---|
Negate | Enable it to mark speech periods. Leave it disable to mark silence |
Time Range | Absolute = Allows you to specify a time interval (in milliseconds) in the audio, during which you wish to get a period of silence Relative = Allows you to specify a time interval (in percent) in the media file, during which you want to search for a period of silence |
Threshold | Specify the duration, in milliseconds, of the silence period you wish to mark. Valid interval: minimum 0, maximum value of 14.400.000. |
Interval | You can explore paragraph below |
Sentiment
This feature underlines the presence or absence of specific sentiments detected in the Call Analytics transcript output. By default the sentiment is analysed on both participants
The rules are designed to match:
The presence or absence of a positive sentiment experienced by both participants at specific points in the call
The presence or absence of a negative feeling felt by both participants at specific points in the call
The presence or absence of a neutral feeling felt by both participants at specific points in the call
The presence or absence of a mixed feeling felt by both participants at specific points in the call
Applicable Fields | Description |
|---|---|
Negate | Enable it to mark sentiments not included in the request. Leave it disable to mark sentiments included in the request |
Time Range | Absolute = Allows you to specify a time interval (in milliseconds) in the audio, during which you want to search for the specified feelings Relative = Allows you to specify a time interval (in percent) in the media file, during which you want to search for the specified sentiments |
Role | Specify the sentiments to be marked among Positive, Negative, Neutral, Mixed (mandatory field) |
Interval | You can explore paragraph below |
Transcript
This rule regards the presence or absence of specific words or phrases detected in the Call Analytics transcript output. A phrase is a set of words separated by space and the words/phrases to be searched are separated by ;
By inserting more words/phrases, the category is activated if it finds at least 1 of the given words
These rules are designed to match:
Custom words or phrases spoken by the agent, the customer, or both
Custom words or phrases not spoken by the agent, the customer, or both (by setting negate option)
Custom words or phrases that occur during a specific period
Applicable Fields | Description |
|---|---|
Negate | Enable it to mark the absence of the sentence in the request. Leave it disable to mark the presence of the sentence in the request |
Time Range | Absolute = Allows you to specify a time interval (in milliseconds) in the audio, during which you want to search for the specified key words or phrases Relative = Allows you to specify a time interval (in percent) in the media file, during which you want to search for the specified key words or phrases |
Role
| Specify the participant you wish to mark (Agent, Customer or Both). Omission of this parameter is equivalent to specifying both participants |
Targets
| Specify the sentences to be marked (mandatory field). It marks the presence or absence of an exact match with the specified phrases. For example, if you specify the phrase “talk to a manager” as Targets value, only that specific phrase is marked. Note that semantic matching is not supported. For example, if the costumer says “talk to the manager” instead of “talk to a manager,” the content is not marked. |
Interval | You can explore paragraph below |
Types of intervals
Start-End
First
Last
You can set Absolute Time Range with interval:
Start - End (e.g. Start 1000 ms - End 12000 ms). So rule to be searched in the inserted interval of voice recordings with start and end
First (e.g. interval from recording start to 12.000 ms)
Last (e.g. interval from 13.000 to the end of recording)
Otherwise you can set Relative Time Range with interval:
Start ( % ) - End ( % ) | (e.g. Start 20% - End 80%)
First (e.g. interval from recording start to %)
Last (e.g. interval from % to recording end)
Related topics