| Add-on | AI agents - Advanced |
With the latest upgrade to the Analytics page we have broken it down into 3 key sections:
Each section is designed to give you insights on your analytics data so you can make data-driven decisions on where the AI agent is performing well, which Intents are bringing you the biggest return on your time investment of setting up that dialogue flow, and which can be improved and are bringing your more escalations.
Our analytics are currently covering data from the last 180 days. We will soon be able to display data from the past 12 months. Reach out to us should you need analytics for a longer period of time.
Before we cover the sections, let's cover the filter criteria that can be defined on the top navigation.
Filters
| Filter | Description |
| Timeframe | This is a date-picker to define the range of dates you would like the analytics and widgets to reflect. |
| Language | A drop-down to filter options based on one or select languages. |
| Conversation Status | This is based on the auto-defined conversation statuses, such as AI agent-handled or escalated. See this article for an explanation of each status. |
| Label | Based on the labels you define, you can filter to show conversations related to them, so this could be Intent-specific labels or related to your A/B testing to compare results. |
| Intent | To have a filtered view, you can drill into one or select Intents to see how they are performing on all metrics. |
| Advanced | Remove test data from the Analytics to have an exact understanding of visitor data, especially useful during any period where you may be testing new flows with the Test AI agent functionality. |
Performance Overview
This section will be the first thing you see and there's a reason for that - it will give you quick insights on key volume metrics focused on understanding, impact, and areas to monitor and potentially troubleshoot. The focus will highlight the differences over time to identify trends either positively based on changes you have made or based on seasonal flows, or negatively which means some work can be done to change the AI agent's performance, same as how you would have a metric dashboard to understand your human agents' performance, this is the same for the AI agent.
| Section | Description | Visual |
| Visitor Volumes |
Understand the number of messages and conversations the AI agent-handled with a focus on understood rate and handling meaningful intents vs just doing a greeting and dropping off. Conversations Fully Understood: Where no message in the communication thread was under the confidence threshold to be not understood.
|
|
| Fully Managed Conversations |
Understand the impact the AI agent has on the success of the support function as it highlights all metrics focused on auto-detected conversation statuses, such as deflection and AI agent-handled rate, as well as the custom resolution rates based on the conversation, flows you have designed and where it indeed reached a positively impactful moment. Use the drop-down to change between AI agent statuses and resolution states |
 |
| Areas to Investigate |
Dive into any issues that have occurred such as failed errors or escalations as well as monitor if the escalation rate is within the normal range or if there is a noticeable change that requires investigation. Use the drop-down to change between escalated conversations and conversations with unsuccessful activity. |
 |
All widgets have drop-down options for you to filter different key information that impacts that respective area. We like to think of the metrics as almost have an equivalent to what you may have within your human customer support team.
- Visitor volumes section is similar to tickets per hour/day/month
- Fully managed conversations is similar to first contact resolution (FCR)
- Areas to investigate is similar to escalations to other teams or where an issue happened that needs a person focused on quality assurance to ensure that everything was handled correctly and where things can be improved.
AI agent performance analysis
This is where your visualization of data takes place - On the left, you can filter for volume-related metrics, and then on the right you have the option of time, language, resolution, and Intent focused metrics.
These are based on the filters you have defined at the top. Then use the drop-down to change between the points that are of most interest to you
Use case performance
At the bottom of the Analytics page, you'll see a list of your use cases and some of their key performance metrics. To sort by any one of them, click either Conversations, Average BSAT, First use case, Custom resolution rate, AI agent-handled rate, Escalations, Failed escalations, or Technical errors to arrange these ascending or descending, or Use case to sort alphabetically by name.
The conversations icon will direct you to the conversation logs for those use case conversations filtered for you.
1 comment
امیر رفیعی
Okay
0