📣 Release: Enhanced Auto QA Metadata Management

Introducing Enhanced QA Scorecard Templates with Auto QA Metadata Management! ⚡️

This update brings a brand-new Auto QA Metadata setup that lets you add extra context and information for your AI-powered quality assessments. With this enhancement, your AI evaluations become even more accurate, targeted, and aligned with your specific QA criteria.

🧩 What’s New

1. A Sleek, Tab-Based Interface

We’ve replaced the old accordion layout with a clean, tab-based design to make setup faster and more intuitive.
Each QA criterion now includes three tabs:

• Criterion Setup – Define your criteria (same as before).

• Auto QA Metadata – Configure contextual data for AI processing (new!).

• Auto QA Instructions – Add guidance for AI evaluations (unchanged).

2. Flexible Metadata Configuration

Customize how your AI understands each criterion with multiple metadata field types, including:
✅ Checkboxes
🔽 Single & Multi-select dropdowns
🧩 Multi-select options
🏷️ Custom tags

All fields are neatly organized into logical groups for easier navigation and setup.

🧩 Auto QA Metadata – Fine-Tune Your AI Ratings

The Auto QA Metadata tab allows you to give your AI more context when evaluating interactions, making ratings smarter, more accurate, and tailored to your workflow.

The metadata configuration is divided into four sections:

1. Platform Fields

Enable standard fields, custom fields, ticket tags, and agent tenure.
These allow the AI to take platform-level data into account while performing ratings, ensuring evaluations reflect key context about tickets and agents.

2. Ticket Content

Include additional ticket-level information like Agent reply time or CSAT survey results.
This ensures the AI considers relevant ticket details when scoring interactions.

3. Advanced Conversation Analysis

Enable advanced options such as reply count or specific keyword detection for deeper insight into agent-customer interactions.
This section allows for more sophisticated and targeted AI evaluations beyond standard metrics.

4. Other

Choose how and what the AI evaluates — for example, whether the agent or end customer should be assessed, or how feedback is structured for evaluations.
This lets you tailor AI evaluation logic to match your business needs.

💡 Pro Tip: You can enable multiple options simultaneously. If you’re unsure which settings are best for your goals, our team can help configure the optimal setup to ensure AI ratings align with your expectations.

When do I need to enable the additional AutoQA Metadata?

You only need to enable AutoQA Metadata when setting up your AI Instructions. It’s not required to select all relevant metadata items before creating the AI Instructions — you can enable them later during the setup process, especially if you realize that your instructions ask the AI to access specific ticket information or perform certain actions.

You can enable multiple metadata items for the same criterion if needed. However, we recommend enabling only the functionalities that are truly required for the AI instructions of the QA criteria.

For example:

• If your QA criteria evaluate adherence to internal processes, the “Agent Reply Time” functionality might be irrelevant.

• Conversely, if you want to evaluate agent reply time, the “Reply Count” functionality is not needed.

Below are examples of AI Instructions that would require additional AutoQA Metadata to be enabled:

---

#1 Example:

Instruction:

Check if the agent closed the chat conversation, while customer was typing their reply. 

In this case, the AI needs to do two additional things besides checking the conversation content:

• “agent closed” — check for a change in the ticket Status.
• “the chat conversation” — focus only on a specific channel type (chat).

Metadata to enable:

#2 Example:

Instruction:

Check if the agent escalated or assigned the ticket to "L2" Group

Here, the AI is being asked to detect a change in the Group assignment, which means it must access the Group field in Zendesk.

Metadata to enable:

#3 Example:

Instruction:

Check if the agent made any grammatical and spelling errors. Do not focus on punctuation or clarity.

This instruction asks the AI to specifically analyze grammar and spelling, regardless of the conversation content.

Metadata to enable:

#4 Example:

Instruction:

When a ticket involves multiple stakeholders (e.g., end customers, suppliers, etc.), and you want the AI to focus on specific responses, use the evaluation target filter to define which replies should be considered for evaluation.

Metadata to enable:

 

For each metadata field, a brief description is provided alongside it to give you additional context on when and how to use it. If you have any specific questions, please don’t hesitate to reach out to us. 🚀