> For the complete documentation index, see [llms.txt](https://guide.tiledesk.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://guide.tiledesk.com/manage-your-workspace/tiledesk-analytics.md).

# Tiledesk Analytics

This guide explains how to read the new Tiledesk Analytics dashboard and understand the main metrics available for conversations, AI Agents, and Knowledge Base performance.

The new Analytics dashboard helps you monitor how your project is performing from one place. You can review conversation activity, understand how much work is handled by AI Agents or humans, check AI token consumption and latency, and evaluate how your Knowledge Base is answering user questions.

<figure><img src="/files/Wygtrh06GffWCegJVHEo" alt=""><figcaption></figcaption></figure>

### How the dashboard is organized

The Analytics dashboard is divided into four main tabs:

#### Overview

Gives a high-level summary of conversations, messages, AI usage, model latency, and Knowledge Base performance.

<figure><img src="/files/Lt6lLK8NFJRkIU8mkPwT" alt=""><figcaption></figcaption></figure>

#### Conversations

Shows detailed analytics about conversation volume, message distribution, and customer satisfaction.

<figure><img src="/files/D7laKLEuzBWhxRtNINyr" alt=""><figcaption></figcaption></figure>

#### AI Agents

Helps you monitor AI Agent activity, token consumption, model latency, webhook triggers, and tool usage.

<figure><img src="/files/RfpcyK1JPStIVNOTGvdu" alt=""><figcaption></figcaption></figure>

#### Knowledge

Shows how your Knowledge Bases are used, how often they answer, and how fast queries are processed.

<figure><img src="/files/iPo0x5d4dYYflmO1yTQL" alt=""><figcaption></figcaption></figure>

### Using filters

Analytics data can be filtered depending on the tab you are viewing. Filters help you focus the analysis on a specific period, department, channel, AI Agent, model, or Knowledge Base.

**Period:** select the time range used to calculate cards and charts.

<figure><img src="/files/sGllFLEkCU5RvWLjiiTi" alt=""><figcaption></figcaption></figure>

**Department:** filter conversation analytics by department when available.

**Channel:** analyze data for a specific channel when available.

<figure><img src="/files/NYAWQAAZp2dhpncRIRC4" alt=""><figcaption></figcaption></figure>

**Agent:** focus AI Agent or Knowledge Base data on a specific AI Agent.

<figure><img src="/files/sUx4Uzh8oZCAEWhaQxIr" alt=""><figcaption></figcaption></figure>

**Model:** analyze token consumption or latency for a specific AI model.

**KB:** filter Knowledge analytics by a specific Knowledge Base.

<figure><img src="/files/lRJtczLVKADeiQRWBeqL" alt=""><figcaption></figcaption></figure>

When you change a filter, the related cards and charts update based on the selected criteria.

### Overview tab

The Overview tab is the starting point of the Analytics dashboard. It gives a quick summary of the most important operational, AI, and Knowledge Base indicators.

<figure><img src="/files/5LLVcBDwIrFCoLBcRlVq" alt=""><figcaption></figcaption></figure>

#### Total Conversations

Shows how many conversations happened during the selected period. Use this card to quickly understand overall conversation volume.

#### Total Messages

Shows the total number of messages exchanged during the selected period. This helps you understand activity level beyond the number of conversations.

#### Average Duration

Shows the average duration of conversations. This is useful for understanding whether conversations are generally short, long, or becoming more complex.

#### Consumed AI Tokens

Shows the amount of AI tokens consumed and indicates how many AI models were used. This helps you monitor AI usage and model consumption.

#### Average Latency Models

Shows the average latency of AI model calls. Use this to understand how fast AI responses are generated.

#### KB Answer Rate

Shows how often the Knowledge Base provides an answer. It also separates answered and unanswered queries, helping you understand how well your Knowledge Base is covering user questions.

#### KB Query Latency

Shows the average latency of Knowledge Base queries. This helps you monitor retrieval performance and Knowledge Base response speed.

### Charts available in the Overview

The Overview tab also includes visual charts that help you understand how conversations, AI usage, model calls and Knowledge Base activity change over time. Each chart includes quick navigation buttons, such as **Go to Conversations** or **Go to AI & Knowledge**, so you can move from the summary view to the detailed analytics section. Most charts also include view controls that let you switch between different visualization formats, such as **Line**, **Area**, **Bar** or **Pie**, depending on the chart.

#### Conversations Over Time

Shows how conversations evolved during the selected period. This chart helps you identify peaks, drops and trends in conversation activity. You can use the **Go to Conversations** button to open the detailed Conversations analytics, and you can switch the chart view between formats such as line, area or bar.

<figure><img src="/files/TNhTjPTT6VHJLt6scd4D" alt=""><figcaption></figcaption></figure>

#### Token Consumption Over Time

Shows how AI token usage changed during the selected period. This chart helps you understand when AI consumption increased or decreased and can be useful for monitoring usage patterns over time. You can use the **Go to AI & Knowledge** button to open the detailed AI and Knowledge analytics, and you can switch the visualization between line, area or bar views.

<figure><img src="/files/ST1OWZG5iFYrJPTxYQFy" alt=""><figcaption></figcaption></figure>

#### LLM Calls Over Time

Shows the number of LLM calls made during the selected period. This chart helps you understand when AI models were called more frequently and how AI activity changed over time. You can use the **Go to AI & Knowledge** button to continue the analysis in the detailed AI analytics section, and you can switch between available chart views such as bar or line.

<figure><img src="/files/K6Smq25b8sYeMaYKcfZa" alt=""><figcaption></figcaption></figure>

#### Model Distribution

Shows how AI usage is distributed across the different LLM models used in the selected period. This helps you understand which models are consuming the largest share of activity. The chart includes a **Go to AI & Knowledge** shortcut, a **Top 10** selector to focus on the most relevant models, and view controls such as pie or bar.

<figure><img src="/files/fICRXNQCE85MFV5P1uDG" alt=""><figcaption></figcaption></figure>

#### KB Queries Over Time

Shows how Knowledge Base queries changed during the selected period. This chart helps you understand when users or AI Agents relied more on the Knowledge Base. You can use the **Go to AI & Knowledge** button to open the detailed Knowledge analytics, and you can switch the chart between line, area or bar views.

<figure><img src="/files/n9SuA3uJhZSnL3vQvqjY" alt=""><figcaption></figcaption></figure>

#### KB Query Latency Over Time

Shows how Knowledge Base query latency changed during the selected period. This chart helps you monitor retrieval performance and identify possible latency spikes. You can use the **Go to AI & Knowledge** button to continue the analysis in the detailed Knowledge section, and you can switch the visualization between line, area or bar views.

<figure><img src="/files/krVZBu9eTcz0IGpcjXR4" alt=""><figcaption></figcaption></figure>

### Conversations tab

The Conversations tab helps you analyze how conversations are created, handled, and rated. It is divided into three sub-tabs: General, Messages, and Satisfaction.

#### General

This section focuses on conversation volume and who handled the conversations.

* **Total Conversations:** shows the total number of conversations created in the selected period.
* **Handled by Agent:** shows how many conversations were handled by the AI Agent, including the percentage of total conversations.
* **Handled by Human:** shows how many conversations were handled by human operators, including the percentage of total conversations.
* **Average Duration:** shows the average duration per conversation.
* **Conversations Over Time:** shows conversation volume over time, split by AI Agent involvement.

Use this section to understand how much work is being managed by AI Agents and how conversation volume changes over time.

<figure><img src="/files/Qy8v9uwTQbIIl7n7teuy" alt=""><figcaption></figcaption></figure>

#### Messages

This section focuses on message volume and message origin.

* **Total Messages:** shows the total number of messages exchanged in the selected period.
* **From Users:** shows how many messages were sent by users.
* **From Humans:** shows how many messages were sent by human operators.
* **From AI Agent:** shows how many messages were sent by AI Agents.
* **Delivered:** shows how many messages successfully reached the inbox.
* **Return Receipts:** shows client-side acknowledgements for messages.
* **Messages Over Time:** shows message volume over time, split by sender type: user, human, and AI Agent.

Use this section to compare user activity, human operator activity, and AI Agent activity.

<figure><img src="/files/xm5Kt1BRrVbuwMazpUR6" alt=""><figcaption></figcaption></figure>

#### Satisfaction

This section focuses on customer feedback and rating quality.

* **CSAT Score:** shows the customer satisfaction score, based on the percentage of ratings equal to or higher than four stars.
* **Average Rating:** shows the average rating received, calculated out of five.
* **Conversations Rated:** shows how many conversations received a rating during the selected period.
* **Positive Ratings:** shows how many ratings were positive, based on four-star and five-star responses.
* **CSAT Trend Over Time:** shows how the CSAT score and average rating changed over time.

Use this section to monitor customer satisfaction and understand whether ratings are improving or declining over time.

### AI Agents tab

The AI Agents tab gives visibility into how AI Agents are being used, how much they consume, how fast they respond, and whether they are executing actions through tools.

#### General

The **General** sub-tab gives a high-level view of AI Agent activity, webhook execution, handovers and action usage during the selected period.

<figure><img src="/files/Qo0hwUX9yffd6JPIuB9w" alt=""><figcaption></figcaption></figure>

At the top of the section, you can use the **Period** and **Agent** filters to narrow the data by time range or by a specific AI Agent.

**Webhook Triggers**

Shows how many webhook triggers were executed during the selected period. This card also indicates how much of the activity was asynchronous, helping you understand how external webhook calls are being used by your AI Agent workflows.

**Unique Agents**

Shows how many different AI Agents were active during the selected period. This is useful when your project includes multiple AI Agents and you want to understand how many of them are actually involved in conversations or workflow execution.

**Sync Triggers**

Shows synchronous webhook executions. These are webhook executions that are handled directly as part of the workflow, where the process waits for the response before continuing.

**Async Triggers**

Shows asynchronous webhook executions. These are webhook executions that can run in the background without blocking the rest of the workflow.

**Agent Distribution**

Shows how intents or interactions are distributed across AI Agents. This helps you understand which AI Agents are handling the most activity. The chart includes controls such as **Top 10**, **Horizontal Bar** and **Pie**, so you can focus on the most relevant agents and choose the visualization that is easier to read.

<figure><img src="/files/FLQ3036cKUNi1qnZyI52" alt=""><figcaption></figcaption></figure>

**Human Handovers**

Shows handovers from AI Agents to human agents. This helps you understand how often conversations are escalated from automation to a human operator.&#x20;

**Action Type Distribution**

Shows how actions are distributed by directive type and AI Agent. This helps you understand which workflow actions are executed most often, such as replies, conditions, task execution, code actions, web requests or other configured directives.&#x20;

<figure><img src="/files/NUG1tKIkuEbflWJg6xtI" alt=""><figcaption></figcaption></figure>

**Action Duration**

Shows the average execution time for each action or directive type, grouped by AI Agent and action type. This helps you identify which actions take longer to execute and can be useful when optimizing workflow performance.&#x20;

<figure><img src="/files/U7qEDzB8KGpE47leTVTD" alt=""><figcaption></figcaption></figure>

**Agent Engagement Over Time**

Shows conversation interactions per AI Agent over the selected period. This chart helps you understand how AI Agent activity changes over time and which agents are more active on specific days. You can switch between **Line** and **Bar** views.

<figure><img src="/files/sAACZR3usrsGz2MrSBTa" alt=""><figcaption></figcaption></figure>

**Action Type Executions Over Time**

Shows how many times each action type was executed over the selected period. This helps you monitor workflow activity and identify which directives are used more frequently over time.&#x20;

<figure><img src="/files/4yV8U9pdoxbORH7HVzcV" alt=""><figcaption></figcaption></figure>

**Agent to Agent Handover**

Shows mid-conversation handoffs between AI Agents over the selected period. This is useful when your workflow includes multiple specialized AI Agents and conversations can move from one agent to another. The section also includes a table with the source agent, destination agent and number of handovers, so you can understand the exact handoff paths.

<figure><img src="/files/DtYxJd1ysb9YzAe65vgM" alt=""><figcaption></figcaption></figure>

**Webhook Triggers Sync vs Async**

Shows webhook trigger activity over time, split between synchronous and asynchronous executions. This helps you understand how external calls to the Tiledesk webhook endpoint are distributed by execution mode. The section also includes a **Webhook Triggers by Agent** table, where you can see total, sync and async webhook triggers for each AI Agent.

Use this section to understand which AI Agents are active, how they interact with users, how often they trigger actions, how quickly those actions execute and how conversations move between AI Agents or human operators.

#### Token Consumption

This section focuses on AI usage and token consumption.

* **Model filter:** allows you to analyze token consumption for a specific AI model.
* **Total AI Tokens:** shows the total number of AI tokens consumed and how many models were involved.
* **Prompt Tokens:** shows input token consumption, meaning the tokens used for prompts and context sent to the model.
* **Completion Tokens:** shows output token consumption, meaning the tokens generated by the model in responses.
* **Total AI Calls:** shows how many AI model calls were made, with a reference to average latency.

<figure><img src="/files/9HvuRWBWKdZwTo844Vdm" alt=""><figcaption></figcaption></figure>

**Token Consumption Over Time:** shows token usage over time, split into total, prompt, and completion tokens.

<figure><img src="/files/XXcBtATO7t1rKJJpABcz" alt=""><figcaption></figcaption></figure>

**Model Distribution:** shows how token usage is distributed across the LLM models used in the selected period.

<figure><img src="/files/XReynn6aiHVuAQLbQnV5" alt=""><figcaption></figcaption></figure>

**Token Usage by Model:** shows prompt and completion tokens for each model. This helps you compare how much each model contributes to overall AI consumption.

<figure><img src="/files/wok9lPddcbNLcl9DrSmR" alt=""><figcaption></figcaption></figure>

**Token Consumption by Source:** shows where tokens are spent, such as RAG, chat or other AI-related sources. This helps you understand which parts of the AI workflow generate the most token usage.

<figure><img src="/files/uKZ9YUflhBA0omwiGIBU" alt=""><figcaption></figcaption></figure>

**Agent Token Ranking:** shows the top AI Agents by token consumption, split between prompt and completion tokens.

<figure><img src="/files/LK5g2Guz8z9VDfHuzWvi" alt=""><figcaption></figcaption></figure>

**Agent Token Consumption Over Time:** shows how token usage changes over time for each AI Agent. This helps you identify which agents consume more tokens during the selected period.

<figure><img src="/files/5kClAyRX67mgPjBb67zM" alt=""><figcaption></figcaption></figure>

**AI Model Calls & Errors:** shows the number of LLM calls and related errors over time. This helps monitor model usage and detect possible issues in AI execution.

<figure><img src="/files/gXqviOPkf6ZuB0olCsPb" alt=""><figcaption></figcaption></figure>

Use this section to monitor AI usage, identify token spikes, and compare how models are being used.

#### Latency

This section focuses on AI performance and response speed.

<figure><img src="/files/wEohmzEiStRTdXi6kGmy" alt=""><figcaption></figcaption></figure>

* **Model Avg Latency:** shows the average latency of AI model responses.
* **Total Model Calls:** shows the total number of calls made to AI models.
* **Model Error Rate:** shows the percentage of model calls that returned errors.
* **Tool Avg Latency:** shows the average latency of tool executions connected to AI Agent workflows.
* **Latency Over Time:** shows how average LLM response time changes over the selected period.
* **Avg Latency by Model:** shows latency grouped by model, useful for comparing performance between models.
* **Model filter:** allows you to isolate latency data for a specific model.

Use this section to check whether AI responses are fast enough and whether a specific model is slower or returning errors.

#### Tools

This section focuses on tool execution, including MCP or internal tool invocations.

* **Total Tool Calls:** shows how many tool calls were made and the related success rate.
* **Successful Calls:** shows how many tool invocations completed successfully.
* **Failed Calls:** shows how many tool invocations returned an error.
* **Avg Latency:** shows the average execution time per tool invocation.
* **Tools Used:** shows how many distinct tools were used during the selected period.
* **AI Tool Calls Over Time:** shows the volume of MCP or internal tool invocations over time.

Use this section to understand when AI Agents are executing actions, how often tools are used, and whether tool calls are successful.

### Knowledge tab

The Knowledge tab helps you evaluate how your Knowledge Bases are used by AI Agents and how well they support answers.

<figure><img src="/files/fEHJUJvMD3W1VyMOSaUm" alt=""><figcaption></figcaption></figure>

#### General

This section gives a high-level view of Knowledge Base usage and answer performance.

* **KBs Active:** shows how many Knowledge Bases were queried during the selected period.
* **Total KB Queries:** shows the total number of Knowledge Base queries across the selected Knowledge Bases.
* **Avg Chunks / Query:** shows the average number of chunks retrieved for each query.
* **Reranking Rate:** shows how many queries used reranking.
* **Indexing Success Rate:** shows the success rate of Knowledge Base indexing operations and whether there were indexing failures.
* **KB Answer Rate:** shows how often the Knowledge Base returned an answer, including answered and unanswered queries.

<figure><img src="/files/X4fRRo8UDNUDoQNqpNji" alt=""><figcaption></figcaption></figure>

**Queries per Knowledge Base:** shows answered and unanswered queries broken down by Knowledge Base.

<figure><img src="/files/4gPt0bcqvwMbEeSUbljj" alt=""><figcaption></figcaption></figure>

**Queries per KB Over Time**: Total, answered and unanswered queries per knowledge base

<figure><img src="/files/jBqwLMwVhxyL6iuNhItU" alt=""><figcaption></figcaption></figure>

Use this section to understand whether your Knowledge Bases are being queried, whether they answer successfully, and which ones may need improvement.

#### KB Latency

This section focuses on Knowledge Base query volume and response speed.

* **KB Avg Latency:** shows the average response time for Knowledge Base queries.
* **Total KB Queries:** shows how many Knowledge Base queries were made in the selected period.

<figure><img src="/files/sDRVBsT0Sb3iO4D63tCM" alt=""><figcaption></figcaption></figure>

**KB Queries Over Time:** shows total and reranked Knowledge Base queries over time.

<figure><img src="/files/kpo0UVOS7FfUnma1Ngbr" alt=""><figcaption></figcaption></figure>

**KB Query Latency Over Time:** shows how Knowledge Base response time changes over time.

<figure><img src="/files/15kV5g6y4rFjaJyDVhm9" alt=""><figcaption></figcaption></figure>

**KB filter:** allows you to analyze latency and query data for a specific Knowledge Base.

<figure><img src="/files/7Tva3Ptz9k02rYhJNbfO" alt=""><figcaption></figcaption></figure>

Use this section to monitor retrieval performance and identify possible latency spikes.

#### Guardrails & Hallucination (**Coming Soon**)

This section is dedicated to future metrics around guardrail triggers and hallucination detection.

Once available, this section will help teams monitor how guardrails are triggered and how hallucination detection behaves in AI Agent workflows.

### What this dashboard helps you understand

The new Analytics dashboard gives teams a more complete way to monitor AI-powered operations. Instead of looking only at conversation volume, you can now connect conversation activity with AI Agent usage, model performance, tool execution, and Knowledge Base behavior.

This makes it easier to understand what is working, where users are getting answers, where humans are still needed, and where your AI Agent or Knowledge Base may need improvement.

We hope you found this tutorial helpful.

To learn more about Tiledesk and how to build AI Agents for support, sales and internal workflows, visit [www.tiledesk.com](http://www.tiledesk.com).

For further assistance, you can contact us at <support@tiledesk.com>.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://guide.tiledesk.com/manage-your-workspace/tiledesk-analytics.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.