About
Welcome to the Conversational Agents Dashboard - your go-to place for understanding how your Kaltura conversational agents are performing. Whether you want a quick snapshot of today's activity or a deeper look at trends over time, this dashboard makes it easy. You can filter by date range to focus on the time period that matters most to you and view key metrics laid out in a clean, easy-to-read format. If you've used other Kaltura Analytics dashboards before, you'll feel right at home. This guide will show you how to get started, explore your data, and make the most of the insights available to you.
Access the dashboard
- Log into the Rich Media CMS and select the Analytics tab from the Content menu.
The Analytics page displays. - In the Analytics menu, select the Agents tab.
The Conversational Agents dashboard displays.
The Analytics page displays.
The Conversational Agents dashboard displays.
Filter by conversation type and agents
- Click the Filters pulldown menu.
- Click to choose a conversation type.
Preview conversations - This is when the admin clicks the Open agent site button on the Overview page to interact with their agent and test the experience. Note that this type of conversation does not consume interactions. See Preview an agent.
Site conversations - This is when the user is having a conversation with the agent on the site and actually consuming interactions. Note that this includes the admin opening the agent site from the Overview page, clicking Open agent site. See Open agent site.
- Click the arrow to expand the Filter by agent pulldown list, then select an agent used during the selected timeframe.
- Click Apply.
Filter by time period
From the date/time pulldown menu, click to choose a period of time for which to filter.
Available data
The first portion of the page provides the key performance metrics at a glance.
Highlights
Threads created - Count of unique thread IDs in the selected time period
Queries sent - Count of all queries across all threads
Avg. queries sent per thread - total queries ÷ total threads
Unique users - Count of unique user IDs that had at least one interaction in the timeframe
Avatar conversations
Number of avatar conversations - count of conversations initiated with an avatar
Total minutes of avatar conversations - sum of minutes over all avatar conversations in selected time period
Average avatar conversation time (in minutes) - total minutes ÷ total conversations
The next portion, Queries over time, focuses on query volume trends over time. This allows you to identify patterns and peak usage periods.
Click to choose time granularity -
- Choose Monthly to display aggregate queries by month
- Choose Daily to display individual days (default)
Hover over data points to see tooltip with exact date and query count
Click View Details below the chart to expand to full detailed view.
The next portion of the page provides the following information:
Response type: Breakdown of answer types provided by the agent, so that you can understand which formats are most commonly used to respond to queries.
Below the chart, you'll find a breakdown (including count and percentage) by query source.
User feedback: User satisfaction feedback visualized so that you can understand the quality of agent responses. On the visual, the red segment indicates bad/negative feedback, and the green segment indicates good/positive feedback. Feedback count, along with percentages of good and bad feedback are shown below the visual.
The final portion provides data on Knowledge sources, including the top sources that the agent referenced, so that you can understand which content is most valuable for generating answers.
The total number of sources that were referenced to generate answers is shown in the upper left corner.
You may enter text in the search box to search for a specific source.
All sources are shown in a ranked list based on number of times they were referenced.
Data includes source name, source type, number of times the source was referenced, and unique users.
You may click on a source link to navigate to the entry dashboard of that source.
Kaltura does not use customer data to train its AI models. To learn more, see Kaltura's Artificial Intelligence Principles.
