Why Do Not Escalate?

When managing large brands or scaling rapidly, a key strategy is minimizing tickets that end up with human support agents. This article provides insights to help you configure your settings effectively and set realistic goals.

Chatform AI offers flexible tools to manage different configurations, catering to the diverse needs and preferences of our users.

Understanding Delegation Rates

Healthy metrics for delegation rates are below 35%, while great metrics are below 30%. Achieving these takes time and involves monitoring conversations, fine-tuning instructions, and filling knowledge gaps.

CS teams often aim to reduce delegated tickets further. While maintaining the goal of providing excellent empathetic player service experiences, resulting in happier players. Let’s explore how this works.

Categories of Delegated Tickets

  1. Knowledge Gaps & Limitations:

    • This is when Chatform AI encounters a limitation where a human might not. This usually stems from knowledge gaps, where Chatform AI lacks instructions to handle specific instances or encounters errors from the back office. Delegation is necessary when Chatform AI isn’t “qualified” to resolve the issue anymore.

    Example Scenario: If a player inquires about the status of their withdrawal, but the backoffice returns an error when Chatform AI attempts to look up the withdrawal.

  2. Proactive Delegation for Insistent Players:

    • Sometimes however, Chatform AI delegates the support ticket to avoid frustrating the player, even when it has followed all instructions correctly. This can occur when a player is insistent or demanding. In these cases, delegating may not add value since the response won’t change. Think of it like contacting support, receiving the correct response, but you are still frustrated with the response so you demand a manager despite them not being able to assist you further as these were the companies policies available.

    Example Scenario: Imagine a player contacts support complaining that they cannt make a withdrawal. Based on your instructions Chatform AI checked the players withdrawal eligibility and informed the player that they are not eligible to make a withdrawal, and explaining the eligiblity requirements. If the player remains frustrated, Chatform AI might typically escalate this ticket to a human agent.

The ‘Do Not Escalate’ Feature

This feature allows you to control category #2 - Proactive Delegation for Insistent Players, situations where Chatform AI provides accurate support but the player remains insistent. By toggling the do not escalate feature, Chatform AI will contain the ticket.

How to setup the ‘Do Not Escalate’ Feature

When you enable ‘Do Not Escalate,’ you’ll find advanced ‘Configure’ option, here you can set a message that Chatform AI will use in these cases. Think of this as a “canned response” from a chatbot where it stubbornly responds with the same message. Here too Chatform AI will use the provided message to insistently respond to the player while maintaining an empathetic tone.

Example Canned Response Message

This is an example of what you might use as a canned response message
Before making a withdrawal, you need to meet the eligibility requirements. Once you meet these requirements, you will be able to make a withdrawal again.

Example Chatform AI Response Message

This is an example of what chatform AI might output based on the above canned
John, I understand your frustration with the eligibility requirements, but unfortunately, these are company policies. Before you can make a withdrawal, you need to meet the eligibility requirements. Once you do, you will be able to make a withdrawal again.

Please let me know if there's anything else I can assist you with.

This approach balances containment and escalation, using empathetic AI to reduce ticket volume while maintaining high service quality.

Beta: Note that this feature is nuanced, and may not be perfect in every case. It may cautiously escalate when it shouldn’t or it may contain even though it would be better to escalate

We’ve fine-tuned this fearure with thousands of examples and ran many tests but we are always seeking to improve our features. If you experience issues or need assistance, please contact our customer support.