Analytics

This section provides an overview of the analytics available for the AI Agent. You’ll find a breakdown of key performance indicators (KPIs), including session volume, user engagement, and response behavior. Additionally, it includes charts and insights on how users interact with the agent across different channels, languages, and time periods. This information is essential for understanding user behavior and monitoring agent efficiency. In Lovi, analytics are organized into four main sections, which can be accessed through the left-hand menu:

• General Metrics: A high-level overview of sessions, users, messages, and agent performance.
• Topic Tracking: Insights into trending topics, FAQs, and automated conversation themes based on user messages.
• Activity Report: Detailed logs of interactions, including timestamps, channels, and response times.
• NPS Report: Net Promoter Score analysis based on user feedback and satisfaction ratings.

These sections allow you to monitor both operational performance and user perception in a centralized, easy-to-navigate dashboard.

​

General Metrics

The General Metrics section allows you to view and analyze your data across various performance indicators and dimensions. It is divided into four tabs:

• Main Dashboard
• Agent Analytics
• Voice Analytics
• Real Time

​

Main Dashboard

The Main Dashboard tab offers a high-level overview of your AI Agent’s activity. It is structured into multiple sections, each focusing on different aspects of the user-agent interaction. You can filter the data by start date, end date, channel, and language to narrow down the analysis and focus on specific timeframes or user segments.

​

Main Dashboard KPIs

This section highlights the core Key Performance Indicators (KPIs) that summarize the agent’s performance and user engagement over a selected time range:

• Total Sessions: Total number of user sessions initiated.
• Unique Users: Number of distinct users who interacted with the AI Agent.
• User Messages: Total messages sent by users.
• AI Agent Messages: Total messages generated by the AI Agent.
• Human Agent Messages: Messages sent by a human agent (if handover occurred).
• Total Messages: Combined number of all messages in both directions.
• Average Messages per Session: Average number of messages exchanged in a session.
• Average Session Duration: Mean length of user sessions, measured in minutes.

These KPIs offer a quick snapshot of how users are interacting with the agent and how the system is performing over time.

​

User Distribution

The User Distribution section provides insight into how users are interacting with the AI Agent across different platforms and languages. It includes two pie charts:

• Messages by Channel: Displays the proportion of messages exchanged through each channel, such as web, voice, WhatsApp, etc.
• Sessions by Language: Shows the distribution of sessions by the language used in each conversation.

By hovering over each segment of the chart, you can view the exact figures and percentages, allowing for quick identification of the most active channels and preferred languages among users.

​

Time Distribution

The Time Distribution section presents a set of visualizations that help you understand when interactions with the AI Agent are taking place and how the agent performs over time. It includes the following charts:

• Session Time Distribution (24h): Displays session volume by time of day. The Y-axis shows sessions grouped by hour, and the X-axis represents the hours in 24-hour format.
• Daily Session Distribution: Shows the number of sessions per day. The Y-axis displays the session count, and the X-axis shows the date in yyyy/mm/dd format.
• Hourly Message Volume: Highlights how many messages are exchanged each hour. The Y-axis shows the number of messages, and the X-axis represents the time of day in 24-hour format.
• Agent Response Time Analysis: Displays the AI Agent’s average response time (in seconds) throughout the day. The Y-axis indicates the response time, and the X-axis shows the hours in 24-hour format.

For all charts, you can hover over each data point to see the exact values and gain deeper insights into user behavior and system responsiveness.

​

Agent Analytics

The Agent Analytics tab focuses on the performance and workload of human agents involved in user conversations. You can filter the data by start date, end date, and agent to analyze specific time periods or agent. It is divided into four key sections:

​

Agent Request Metrics

Displays high-level data on how sessions are escalated and handled by agents. Metrics include:

• Assigned to Human
• Assigned Automatically
• Assigned Manually
• Agents Involved
• Agent Timeout Closures
• User Timeout Closures
• Average Session Time with Agent
• Quick Responses (under 5 minutes)

​

User Distribution

Highlights the distribution of escalations by:

• Channels with Highest Escalations
• Languages with Highest Escalations.

This helps identify where most handovers to human agents are taking place.

​

Agent Detailed Metrics

Provides agent-level performance data, including:

• Chats Assigned by Agent
• Average Chat Time per Agent (minutes)
• Chats Closed per Agent
• Total Connection Time per Agent
• Agent Availability (hours). Different agent statuses are represented with distinct colors: Online, Break, Ending Shift, and Training. This visualization helps you clearly identify each agent’s availability throughout the day.

​

Agent Time Distribution

Visualizes the distribution of agent activity over time:

• Agent Time Distribution by Hour
• Agent Time Distribution by Day.

These charts help you understand agent coverage and workload trends. Each chart allows you to hover over the data points to view exact values for more precise analysis.

​

Voice Analytics

The Voice Analytics tab provides insights into the performance and quality of voice interactions handled by your AI Agent. You can filter the data by start date and end date to focus on specific timeframes. These metrics are essential to evaluate both user experience and system performance in voice-based interactions. This tab includes the following sections:

​

Voice KPIs

Displays key performance indicators related to voice calls, including:

• Total Calls
• Average Call Duration (min)
• Total Call Duration (hours)
• Successful Scored Calls
• Success Rate (%)
• High Stress Calls (%)
• Average Talk Duration (seconds)

​

Average Score Distribution

Visualizes the distribution of average scores assigned to calls, helping assess overall call quality.

​

Call Trends

Shows how voice call volume and performance evolve over time, helping you identify patterns, spikes, or drops in activity.

​

Real Time

The Real Time tab provides a live snapshot of ongoing activity within your AI Agent ecosystem. It helps monitor operational dynamics and agent performance in real time. The data can be used to make immediate decisions and optimize live support. This tab includes two main sections:

​

Live Metrics

Displays real-time figures for:

• Users Assigned to Agents
• Users in Queue
• Agent Workload
• Bot to Human Transfer Rate
• Average Wait Time (min)

​

Agent Status

Lists individual agents with the following information:

• Name
• Status (e.g., Online, Offline)
• Current Workload
• Languages each agent is assigned to handle

This section is especially useful for supervisors who need to track availability and workload at a glance.

---

​

Topic Tracking

The Topic Tracking section (formerly Social Listening) allows you to automatically track, measure, and categorise what your users are talking about. Instead of manual sorting, your AI Agent does the heavy lifting by analysing conversations and grouping them into a clear hierarchy: TOPIC ➔ FAQ ➔ LINKED MESSAGE This section is divided into two main areas in the left-hand menu: Tracking and Conversation Insights.

​

⚙️ How the AI Processing Works

The system operates in smart batches. Every time your AI Agent receives 50 new user messages, it triggers an automated processing cycle. The AI analyses these 50 messages, reads the instructions you provided in the Prompt Topic (located in your Agent Settings), and automatically generates relevant Topics and FAQs, linking the specific user messages to them.

⚠️ Crucial Note: If the Prompt Topic field in your Agent Settings is empty, the AI will not process messages or create any Topics/FAQs.

⭐ FAQ Ratings: You can also configure an “FAQ Rating” prompt in your Agent Settings. When active, if a user asks a common question (e.g., “What time does the mall open?”), the AI will reply with the existing FAQ and append a quick survey (e.g., “Did you find this helpful?”). When the user leaves a rating, this feedback is counted and reflected in your metrics during the processing cycle.

​

📊 Tracking

This is your main dashboard for visualising and managing your categorised data. It contains three tabs:

​

Topics Analytics & FAQs Analytics

These tabs provide a visual summary of your data based on the links made by the AI. You can view:

• Topic Distribution: A chart showing how mentions are spread across your broader topics.
• Top Topics by Mentions: A bar chart displaying the exact number of times a topic was referenced.
• FAQ Performance: See how frequently specific FAQs are triggered and review the ratings left by users.

​

Tracking Management (Manual Control)

While the AI works automatically, you remain in full control. From this tab, you can manually create, edit, and delete your Topics and FAQs.

• The Deletion Rule: The system protects your data. If you try to delete a Topic or FAQ that already has user messages linked to it, you won’t just lose the data. The system will prompt you to either reassign those linked messages to a different Topic/FAQ or permanently delete them alongside the category.

​

🔍 Conversation Insights

Sometimes, messages cannot be processed automatically (due to system glitches, unclear context, etc.). The Conversation Insights menu provides a safety net for these situations.

​

Unclassified Messages

This section displays a list of messages that failed to process. Instead of leaving them in limbo, you can use the refresh/reprocess button to force the AI to analyse these specific messages again and properly categorise them into your Topics and FAQs.

---

​

Activity Report

The Activity Report provides a detailed log of recent user interactions with your AI Agent. It includes key information such as the language used, the communication channel, the timestamp of the session, and whether the conversation was escalated to a human agent. You can filter the report by start date and end date, and also download the data as a CSV file for further analysis. The table includes the following columns:

• User: Identifies the user involved in the interaction.
• Human Escalation: Indicates whether the session was escalated to a human agent.
• Metadata: Displays the language of the conversation and the channel used (e.g., web, voice, WhatsApp).
• Created: Shows the date and time when the session was initiated.

This section is useful for auditing interactions, monitoring escalation patterns, and analyzing user behavior over time.

​

NPS Report

The NPS Report section allows you to measure user satisfaction by setting up and analyzing Net Promoter Score (NPS) surveys. NPS is a widely used metric that helps you understand how likely users are to recommend your service, based on their experience with your AI Agent or human support. You can filter results by start date and end date, and create new NPS surveys using the Create NPS Survey button. When creating a new survey, you can configure the following fields:

• Channels: Select which communication channels (e.g., web, WhatsApp, voice) the survey will apply to.
• Conversation Type Policy: Define the type of conversation that should trigger the survey. The available options are:
  • skip_if_agent: the survey will be skipped if the conversation involved a human agent.
  • only_if_agent: the survey will only be sent if a human agent participated in the conversation.
  • none: the survey will be sent regardless of human involvement.
• Delay Minutes: Specify how long to wait before sending the survey after the interaction ends.
• Enter Flow: Choose the flow or path the user will follow when completing the survey.
• Message Mapping: Map each step of the survey to specific messages that will be shown to the user.

This section is ideal for continuously collecting feedback and identifying areas for improvement in your customer service experience.