SAC 360º - AI Analysis - Evaluating services automatically and intelligently

With the new AI Analysis feature, it is now possible to automate the evaluation of support interactions as soon as they are closed. The AI analyzes the interactions, generates improvement suggestions based on defined criteria, and even notifies managers when it identifies low scores.

This feature makes it possible to monitor service quality in a faster, more scalable way with less manual effort.

This new feature is available only to those who have purchased the AI module.

How does AI Analysis work?

As soon as an interaction is closed, the AI system kicks in, processing the messages and generating an evaluation based on the criteria configured by you. The analysis can be guided by several filters and parameters.

1. Choose the analysis model

You will be able to select which AI agent model you want to use for the evaluation. This choice defines how the AI will interpret the conversations and apply the defined criteria.

2. Limit analysis by user

If you want the AI to evaluate only interactions handled by certain users (e.g., new team members), you can select those users. Leave the list empty if you want to apply the analysis to all team users.

3. Limit analysis by account

You can also restrict the analysis by support channels, such as WhatsApp, Facebook, or Instagram. This is useful for testing the AI on specific channels or in more strategic interactions.

Define the evaluation criteria

The evaluation criteria are the parameters the AI will use to score each interaction. You can:

• Define the criteria that should be evaluated.

• Add descriptions to make the criterion clearer to the AI.

• Apply specific criteria to selected accounts.

The final score will be based on the overall judgment across all criteria, and the AI can also highlight positive or negative points in performance reports.

Manager Notifications

You can configure automatic notifications for managers or team leaders whenever an interaction receives a rating below a specific score (from 0 to 10). To do this, you can define:

• Who will be notified (Manager, CEO, Administrator, etc.).

• Which users will be monitored.

• What minimum score will trigger the alert.

Additional settings

In addition to the filters and criteria, the feature allows extra adjustments to refine its use:

• Add analysis log to the interaction: if enabled, the analysis result will be visible directly in the support conversation.

• Minimum messages for analysis: defines the minimum number of messages for the AI to perform the analysis.

• Maximum messages for analysis: defines the maximum number of messages per interaction, avoiding very long and costly analyses.

View the results with charts and full reports

After the analyses performed by the AI, you can track all the results through interactive charts and detailed reports, which make it easier to understand team performance and help with decision-making.

This data can be filtered by:

• Account: WhatsApp, Facebook, Instagram, or Live Chat.

• Owner: Name of the user who handled the interaction.

• Period: Refers to the interaction closing date, allowing analyses by day, week, month, or quarter.

Charts - Conversation Analysis

In the charts area, you will have access to key indicators about the analyzed interactions:

• Total Evaluations: Total number of analyses performed in the selected period.

• Average Score: Overall average of all completed evaluations.

• Positive and Negative Highlights: Criteria most frequently identified as strengths or weaknesses in the conversations.

• Average Participation: Shows how much the user actively participated in the interactions. Example: of the messages sent, 65% were from the responsible user.

Chart Table

Below the main indicators, there are detailed tables with the following information:

• Positive Criteria: Chart with the distribution of the criteria that performed best in the evaluations.

• Negative Criteria: Distribution of the criteria that need the most attention and improvement.

• Average Score by Account: Comparison of the average score by channel (WhatsApp, Facebook, etc.) during the analyzed period.

These charts help you understand, for example, whether Instagram interactions perform better than WhatsApp, or which criteria are most recurring in each channel.

Detailed Reviews

In addition to the charts, the system also provides a complete list of all analyses performed. In each review you will find:

• Lead ID and Interaction: Interaction identification (with a direct view link).

• Account and Owner: Channel used and who handled the interaction. Score: Rating assigned by the AI based on the criteria.

• Evaluation Date.

• Total Messages: Total number of messages exchanged in the interaction.

• User Messages: Total number of messages sent by any team user.

• Owner Participation: Percentage of messages sent specifically by the responsible agent.

• Positive and Negative Criteria: Which criteria were highlighted by the AI.

• Model Used: AI model that performed the analysis.

• Input/Output Tokens: Number of tokens processed, useful for technical and cost control.

In addition, it is possible to export all reviews in a .CSV file, making external analyses, presentations, or audits easier.

Conclusion

The AI Analysis feature was created to optimize quality management in customer support, making the evaluation process more practical, intelligent, and data-driven. With it, you ensure constant monitoring of the team and can act quickly on interactions that require attention, all in an automated way.