Voice of Customer and Quality Assurance

The analysis of Voice of the Customer (VOC) and Quality Assurance (QA) tools in Wolkvox are centralized in the Voice and Text Analysis and Quality modules.

1. Quality Assurance (QA) Functionalities

The Wolkvox platform offers tools for the structured evaluation of agent performance through quality matrices and specific reports:

1.1. Evaluation and Monitoring Tools

• Sending Calls to Quality Analyzer: Supervisors or quality analysts can send specific call recordings from Data Monitor to the Quality Analyzer for detailed quality analysis. This action complements the automatic selection of calls made by the system.
• Quality Matrix: Service quality is evaluated using the Quality Matrix, which defines performance attributes that can be classified as critical (CE) or non-critical (NCE). The total sum of the percentages assigned to each attribute must be 100%.
• Wolkvox QA AI Dashboard: Located in the "Wolkvox QA AI" tab within "Voice and Text Analysis" on the Dashboard, it allows analyzing the accuracy percentage in meeting the attributes defined in the Quality Matrix.
  • Analysis Views: Information can be consulted by Agent, by Attribute, or by Channel (e.g., chat-WhatsApp, chat-web, voice).
  • Visualized Metrics: Shows the number of interactions analyzed and the accuracy percentage.

1.2. Detailed Quality Reports (Quality Tab in Reports)

Reports in the "Quality" tab allow for a thorough audit of the evaluation process:

2. Voice of the Customer (VOC) Analysis and Speech Analytics

These functionalities focus on analyzing the content of interactions (voice and text) using artificial intelligence to measure sentiment and extract key information.

2.1. Detailed Analysis per Interaction (CDR Speech)

The "1. CDR Speech" report (in the "Voice Analysis" tab) provides a comprehensive analysis of interactions:

• Identification and Times: Includes AGENT_ID, CONN_ID, TIME (duration), DATE, CUSTOMER_PHONE, and COD_ACT.
• Voice-to-Text Analysis (Transcription): Contains the AUDIO_TEXT field, which is the transcription of the audio conversation to text.
• Customer Sentiment: The FEELING (Sentiment) field classifies the customer's emotional tone into five values: very negative, negative, neutral, positive, and very positive.
• Silence Detection: The SILENCE field shows the number of seconds of silence in the conversation (only applies to the "voice" channel).
• Agent Greeting Analysis: Shows whether the agent greeted (GREETING) and if it was slow (SLOW_TO_GREET). It also indicates if the agent mentioned their name (GREETING_AGENT_NAME).
• Custom Categories: Allows up to 10 configurable columns to identify keywords or specific phrases within the conversation.
• Supported Channels: The analysis applies to channels such as voice, chat-whatsapp, chat-web, chat-facebook, chat-instagram, and chat-sms.

2.2. Content Visualization and Monitoring

• Topics: A visual representation from the Dashboard ("Voice and Text Analysis"). Shows the most frequent situations used in interactions (generated from transcriptions or chats).
• On-Demand Analysis (Speech on Demand): The "5. Speech on Demand" report allows analyzing specific conversations to obtain a summary (SUMMARY) and the identified average sentiment (FEELING).

2.3. Agent Vocal Performance Reports (Voice Analysis Tab)

There are specific reports to measure the agent's vocal behavior:

3. AI (NLP) Component Functionalities

Voice and text analysis functionalities are supported by Natural Language Processing (NLP) components:

4. Smart Survey Functionalities (VOC)

The platform uses smart surveys to obtain structured customer feedback:

• Monitoring in Data Monitor: Satisfaction surveys can be searched and monitored in the "IS Survey Inbox" tab of Data Monitor. The table shows the ratings of the questions (Q01 to Q10), duration, and Sentiment.
• Survey Reports (Diagram Reports):
  • Smart Survey Detail: Report that offers detailed answers to each question (Q01-Q10), the duration of the voice message if the customer left one, the result (abandon, complete, incomplete), and the customer's sentiment (FEELING).
  • Smart Survey by Agent: Consolidates the number of surveys conducted for an agent (SURVEYS) and the average rating obtained by the agent for each question (Q01-Q10) and the overall average (AVERAGE).
  • Smart Survey by Skill: Consolidates the number of surveys by skill and the average ratings per question (Q01-Q10) and the overall average for the skill.

Speech Analytics

Voice and text analysis in Wolkvox is supported by Speech and Text Analytics components that analyze the content of interactions to extract key communication and sentiment metrics.

A. Key Components and Metrics

Automatic Quality Assurance (Auto QA)

The QA functionality in Wolkvox focuses on measuring the accuracy percentage in meeting the attributes defined in the Quality Matrix. The system uses AI to perform automatic evaluations.

A. QA Process and Configuration

• Quality Matrix: Quality is evaluated using this matrix, where performance attributes are defined.
• Error Classification: Attributes are classified as Critical (CE) or Non-Critical (NCE).
• AI Quality Analyst: If the evaluation was automatic using artificial intelligence, the QUALITY_ANALYST column in the reports will show the value QAi_Bot.
• Manual Sending to QA: Supervisors can select specific call recordings in Data Monitor and manually send them to the Quality Analyzer (Send audio to QA) to complement the system's automatic selection.

B. Wolkvox QA AI Dashboard

The Dashboard, under the "Wolkvox QA AI" tab (within "Voice and Text Analysis"), centralizes the analysis of quality results.

• Analysis Views: Information can be consulted and segmented by:
  • By Agent.
  • By Attribute.
  • By Channel (e.g., chat-WhatsApp, chat-web, voice).
• Metrics Displayed: Shows the number of interactions analyzed and the accuracy percentage in meeting the attributes.

C. Detailed Quality Reports (Quality Tab in Reports)

1. Results by Agent:
   • Shows the consolidated performance of the agent, including the number of Evaluations, Accuracy in Critical Errors (PRECISION_UNIT_CRITICAL_ERROR, PRECISION_OPPORTUNITY_CRITICAL_ERROR), and Accuracy Percentage (ACCURACY).
2. Result by Evaluation by Agent:
   • Offers the detail of each individual evaluation applied to an agent. Includes CONN_ID, AGENT_NAME, COMMENTS, accuracy and precision metrics, DATE, CHANNEL, and the name of the analyst (QUALITY_ANALYST).
3. Evaluation Detail by Attribute:
   • Allows you to see whether each individual attribute evaluated was met (Yes) or not (No). Details the ATTRIBUTE, its classification (CE for Critical or NCE for Non-Critical), and who performed the evaluation (QUALITY_ANALYST or QAi_Bot).
4. Results by Attribute:
   1. Consolidates the global compliance of each ATTRIBUTE in the matrix. Shows the number of times it was met (YES) or not (NO), and the average compliance percentage (PERCENT).