> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reddy.io/llms.txt
> Use this file to discover all available pages before exploring further.

# QA Scorecard Configuration

> How to set up your QA items so they are automatically graded by Reddy's AI

## Overview

QA scorecards define the criteria that Reddy's AI uses to automatically evaluate and grade calls. Each QA item consists of a title, type, requirement description, point value, and status indicator. These elements work together to power Reddy's automatic grading and agent feedback.

<Info>
  Need step-by-step setup instructions? See the [QA Setup Guide](/manual/qa/qa_setup) for detailed configuration steps.
</Info>

## Understanding QA Item Elements

<Tabs>
  <Tab title="Type Options">
    <AccordionGroup>
      <Accordion title="Pass/Fail">
        **Best for**: Binary requirements where agents either meet the standard or don't

        **Examples**:

        * "Agent provided account verification"
        * "Agent used proper greeting"
        * "Agent followed closing script"

        **Scoring**: Simple pass or fail determination
      </Accordion>

      <Accordion title="Score">
        **Best for**: Requirements where performance can vary in quality

        **Examples**:

        * "Agent demonstrated empathy" (1-5 scale)
        * "Agent provided accurate information" (1-10 scale)
        * "Agent resolved customer issue effectively" (percentage-based)

        **Scoring**: Numerical grade based on performance quality
      </Accordion>
    </AccordionGroup>
  </Tab>

  <Tab title="Requirements">
    The requirement description is where you define what the AI should evaluate. Reddy transforms your written requirements into automatic grading logic.

    <AccordionGroup>
      <Accordion title="Writing Effective Requirements">
        * Be specific about expected behaviors, phrases, or actions
        * Include context about when the requirement applies
        * Define success criteria clearly
        * Use examples to illustrate compliance
      </Accordion>
    </AccordionGroup>

    **Example Requirements:**

    <Tabs>
      <Tab title="Greeting">
        **Weak**: "Agent greets customer"

        **Strong**: "Agent greets customer with: company name, their own first name, and offers assistance. Example: 'Thank you for calling \[Company], this is \[Name], how can I help you today?'"
      </Tab>

      <Tab title="Verification">
        **Weak**: "Agent verifies account"

        **Strong**: "Agent verifies the customer's identity by collecting at least two of the following: full name, phone number, email address, or account number before discussing account details."
      </Tab>

      <Tab title="Empathy">
        **Weak**: "Agent shows empathy"

        **Strong**: "Agent acknowledges customer frustration or concern with empathetic language such as 'I understand how frustrating that must be' or 'I can see why you'd be concerned about this' before moving to resolution."
      </Tab>
    </Tabs>
  </Tab>

  <Tab title="Points">
    Point values are weighted relative to all other items in the scorecard. Higher point values increase an item's impact on the overall compliance score.

    <Tip>
      Align point values with your organization's priorities. If account verification is critical, assign it more points than closing pleasantries.
    </Tip>
  </Tab>

  <Tab title="Status Indicators">
    **Green Status**: QA item is active and being automatically graded by Reddy's AI

    **Yellow Status**: QA item needs AI model updates before automatic grading activates. Contact your Reddy team if items remain yellow for extended periods.
  </Tab>
</Tabs>

## Organizing Your Scorecard

Group related QA items into sections that follow your call flow. This makes scorecards easier to navigate and ensures evaluations track the natural progression of customer interactions.

**Example:**

* **Introduction** (section) contains: Greeting, Agent identification, Confirm call purpose
* **Closing** (section) contains: Confirm next steps, Thank customer

Use drag-and-drop to reorder items within and across sections.

## Fine-Tuning Your Scorecard

After configuring your QA scorecard, test it against real customer calls to ensure Reddy's grading accuracy matches your quality standards.

Start with 20 manually scored calls to establish a solid benchmark dataset. Focus on calls representing typical scenarios.

### Prerequisites

Before evaluating QA scorecard accuracy, upload real customer calls to test your scorecard against actual interactions. Use these calls to verify that Reddy's AI grades match your quality standards and to identify any scoring logic that needs refinement.

[Upload Call Recordings](/manual/qa/manually_uploading_calls)

### Evaluation Process

Review AI-scored calls and flag discrepancies to refine Reddy's QA logic. This feedback improves scoring accuracy over time.

[QA Evaluations Guide](/manual/qa/qa-evaluations)