Branching, free text and escalation
Who it is for: whoever manages the account
TalkAI surveys are not linear: the path can change depending on what each person answers.
Branching by option
Each answer option (or each free-text rule) can have an action:
| Action | What it does |
|---|---|
| Continue | Moves to the next question |
| Jump to a question | Goes to a specific question |
| Escalate | Hands the conversation to a person or an AI agent |
| End | Finishes the survey early |
Example: for "Are you happy with the service?", if they answer "No" you can jump to a question asking for details, or escalate to a person.
Rules for free-text answers
When the question is free text, there are no buttons: TalkAI interprets what the contact writes with content rules. Each rule defines a pattern and an action.
- Contains: matches if the answer includes certain text (for example "not working" â escalate). Case-insensitive.
- Regular expression: for advanced cases (validating a code format or a number).
Rules are evaluated in order: the first one that matches decides the path.
You can also require a valid format with a regular expression: if the answer does not comply (for example, it is not an email), it counts as invalid and is retried or escalated.
Escalation to a person or the AI
A survey goes from automatic to handled in these cases:
- The contact picks an option with the Escalate action.
- The answer exceeds the maximum number of invalid answers.
When escalating, you choose the routing:
- Human: a conversation is created in the department queue (the option's, the campaign's or the default one). Someone picks it up from the Inbox.
- AI: it is handed to a specific AI agent, which continues the conversation.
- Automatic: the system decides (by default, it hands off to a person).
When escalated, the contact gets a notice that their request was forwarded, and the conversation appears in the Inbox ready to answer.