Intelligent email routing
Pega Intelligent Virtual Assistant (IVA) for Email uses routing conditions that you define to assign a customer email to an operator, route it to a work queue, create a top-level regular case of case type, or send an automatic email reply. For example, a customer email about a problem with a credit card transaction can be automatically routed to an operator, and a regular case created to help track the issue.
Intelligent routing mode
When you define routing conditions for actions during intelligent routing configuration, you must also indicate to process an email based on:
- First matching action - The system triggers only the first action that is defined from the top of the list that matches conditions. All other actions in the list with matching conditions are not triggered.
- All matching actions - The system triggers every create case and automatic reply action in the list that matches conditions while all route to actions are skipped. If no create case or automatic reply action matching conditions are found in the list, the system triggers only the first route to action that matches conditions.
Routing conditions
You can define routing conditions for an action that are based on the following criteria:
- Email from (both email and name) header
- Email to header
- Email subject header
- A natural language processing (NLP) entity that is extracted from the content of the email using text analysis
- A topic that is inferred for the content of the email using NLP text analysis
- A sentiment value between 0 and 1 that is calculated based on the content of the email using NLP text analysis
- A localized language that is detected using NLP text analysis
- A when rule
If a routing condition holds true, then its action is automatically triggered in the application, for example, a top-level regular case is opened with the received email content.
You can set up any number of routing conditions for intelligent email routing. However, in the Otherwise area on the Behavior tab you must specify what to do when:
- By default, no routing conditions were defined.
- None of the defined routing conditions hold true.
If none of the routing conditions hold true in the list, then the rule specified in the Otherwise area is applied.
Email content text analysis
The text analysis settings for an Email channel interface consist of entities, topics, sentiment, and language. These settings provide sentiment analysis, text (topic) classification, intent detection, and entity extraction of text-based content for the received email. The settings are used in routing conditions. These routing conditions help the system determine whether to assign the received email content to a user, add it to a work queue, or whether a top-level case is created with the information. For example, the location entity that is defined in a routing condition helps the Email channel determine where to route the email based on the value extracted for location from its content.
Default settings from a text analyzer rule are used for each Email channel interface. You can configure the text analyzer topics directly from the Text analysis tab so that the Email channel can correctly recognize responses in received emails. You can also edit the text analyzer rule that defines the natural language processing (NLP) model for the Email channel, which includes the sentiment analysis, text (topic) classification, intent detection, and entity extraction settings.
Topic confidence threshold
When you define a routing condition with a topic, you can also specify the confidence score threshold with which to decide whether a topic matches the input text, that is, the content of the received email. The confidence threshold is used in text analyzer classification analysis and gives you more control over which text content is assigned to a topic. For example, in a routing condition you can specify that a topic is determined to be a greeting message only with a confidence score of 0.8 or higher.