Quality Analysis: Categories
On this page |
FROM VERSION 3.36.0
Overview
Categories are collections of filters used in Post Call Analytics to analyze voice recordings. Each category contains a set of rules that define specific conditions to detect during transcript analysis.
Categories must be created before submitting a transcript request. They cannot be applied retroactively.
By default, all created categories are automatically applied to transcripts generated through Call Analytics.
Each category can contain 1 to 20 rules. Rules define what conditions trigger category detection. Categories are associated with a specific cloud provider account and region.
In the QA Categories section, under Voice menu, you can view all created categories along with their associated cloud provider account and region.
FROM VERSION 3.56.0
A new user experience is available for the QA Categories section.
Use the three dots menu next to a category to edit or delete it.
Categories can be sorted by name, filtered by region, and display columns can be customized.
Additionally, you can select multiple categories via checkboxes to delete or export them using the available action buttons.
Each category includes:
Category Name
Associated Cloud Provider and Region
Creation and Last Update Timestamp
Requirements
QA Categories are available from XCALLY version 3.36.0
The New experience for the QA categories section is available from version 3.56.0
You need the Quality Analysis add-on license
QA Categories are supported only on AWS Cloud Provider
Video Tutorial
Creating a new Category
To create a new QA Category:
Click the Add button
Provide the following details:
Cloud Provider: select from previously configured accounts
Region: automatically populated based on the Region selected in Region filter (QA Categories view)
Name: define a name for the category
The category will be created with a default rule, which can be modified later.
Click on Add to complete the creation
To add or modify rules:
Click the three dots next to the category and select Edit.
Go to the Rules section.
Click Add to create a new rule.
Choose a Rule Type and fill out the necessary fields.
Rule Type: each rule defines a specific condition for analyzing voice recordings. When multiple rules are configured in a category, all rules must be satisfied for the category to be triggered. Select among:
Interruption time
Non-talk time
Transcript content match
Transcript sentiment match
Fill the fields according to the type of rule you selected
Click the Confirm button
See the next paragraph “Types of Rules” below for a full description of each rule type.
Types of rules
You can configure 4 types of rules and if you create multiple rules, all must be met to track the desired elements within the voice recording.
Interruption
This rule detects the presence of interruptions in the Call Analytics transcript. It helps evaluate whether the agent interrupts the customer, the customer interrupts the agent, or neither.
You can configure:
Interruption Duration: Minimum duration (in seconds) to qualify as an interruption, so when the duration of the interruption is more than the time indicated here
Time range to consider → during the:
entire call
first: defines the interval from the beginning of the recording up to the specified time (e.g. the first 12 seconds)
last: defines the number of last seconds to consider (e.g. the last 13 seconds of the call)
Speaker, that can be:
Agent, which interrupts customer
Customer, which interrupts agent
Either participant interrupts the other (default option)
Non-Talk Time
This rule tracks the presence of silent periods exceeding a set threshold during the call. While normal pauses are expected, longer silences may indicate holds or issues.
You can configure:
Silence duration (in seconds), so when no word has been spoken for more than the time indicated here
Time range to consider → during the:
entire call
first: defines the interval from the beginning of the recording up to the specified time (e.g. the first 12 seconds)
last: defines the number of last seconds to consider (e.g. the last 13 seconds of the call)
Transcript sentiment match
This rule marks the presence of specific sentiments detected in the transcript, applied by default to both participants.
You can configure:
Sentiment type: Positive, Negative, Neutral, or Mixed sentiments present during specific parts of the call.
Time range to consider → during the:
entire call
first: defines the interval from the beginning of the recording up to the specified time (e.g. the first 12 seconds)
last: defines the number of last seconds to consider (e.g. the last 13 seconds of the call)
Speaker, that can be:
Agent
Customer
Either (default option)
Transcript content match
This rule detects the presence or absence of specific words or phrases in the transcript. Type one or more words or phrases and press Enter to add it. The rule triggers if at least one phrase is found.
You can configure:
if you want to trigger the rule when the words are mentioned or not mentioned
Time range to consider → during the:
entire call
first: defines the interval from the beginning of the recording up to the specified time (e.g. the first 12 seconds)
last: defines the number of last seconds to consider (e.g. the last 13 seconds of the call)
Speaker, that can be:
Agent
Customer
Either (default option)
Specify exact words or phrases to detect (mandatory). You need to insert minimum one element.
Example: “talk to a manager” will match only that specific phrase.
Semantic variations are not recognized. For example, if the costumer says “talk to the manager” instead of “talk to a manager,” the content is not marked.
Once rules are added to a category, they will appear in the Rules section of the category editor.
You can edit or delete individual rules using the three dots menu next to each rule.
Related topics