Tags are one of the main components of Echo AI. It is a powerful feature that helps provide the insights into specific conversations that you are looking for and even some that you didn't even realize you needed. These are done through Configured and Generated Tags.
Configured vs. Generated Tags
While Configured and Generated Tags both aim to categorize conversations effectively, their setup and application differ. Generated Tags are the AI's spontaneous response to emerging sentiments and themes within conversations without needing prior configuration. In contrast, Configured Tags are based on predefined criteria you set up in the Admin settings. They're specifically crafted to detect conversations based on user-defined themes or topics.
Generated Tags
The Customer Intelligence (CI) Generated Tags feature automatically detects and labels evolving trends, prospects, and potential challenges in conversations. Crucially, these tags spotlight previously unrecognized themes within your dialogues.
How does it work?
When the Large Language Model (LLM) evaluates that a conversation revolves around a specific topic or issue or notices a possible trend, it creates an appropriate tag for that interaction. For instance, if a discussion touches upon a compliance concern, the LLM will automatically assign a relevant tag, such as "Compliance Issue" to the conversation.
Benefits of the Feature:
Generated Tags offer a slew of advantages:
Hands-Off Insights: You don't just have to configure a tag to uncover common trends or themes, as CI will automatically create tags for you on what it sees ensuring that you always know what is happening in your conversations.
Spot Emerging Trends: Easily recognize patterns, allowing for more informed decision-making.
Uncover Hidden Risks: Proactively detect potential issues like fraud or compliance breaches.
Understand Your Customers Better: By immediately grasping the core themes of conversations, you can more deeply understand and respond to the Voice of the Customer.
Configured Tags
The CI Configured Tags feature is designed to label and categorize conversations based on specific, company-specific criteria. This enhances the clarity and organization of interactions and sets the stage for deeper analysis and action.
Types of Configured Tags
Simple (Phrase Matching) Tags - These are configured tags that look for a keyword or phrase. If the keyword or phrase is found in a conversation, the tag will apply.
Prompt (LLM) Tags - These tags require a prompt, a description of the specific case that guides the LLM as to which conversations this tag should apply to. These can be complicated and nuanced to configure.
Multiple Choice Tags - Under Multiple Choice Tags there are two types:
Binary (always choose one option) - There are only two choices and only one of the options can be chosen.
Multi Select (choose at most all options) - There can be many choices and anywhere from none of them to all of them can be chosen.
How do I make an effective Configured Tag?
You can configure a configured tag using a natural language prompt in pipeline settings. For example, for a tag named "Login issues" that prompt might be "The customer had trouble logging in or accessing their account. Configured Tags allow you to choose who the Configured Tag should be analyzed for Customer, Agent, or Either. You can also assign a high priority or low priority to the tag. High priority tags are listed first and have more prominent styling. Low priority is assigned to the tag by default if the high priority option is not toggled on (see the image below).
Configured Tag example:
Tag | Access Issues |
Tag Prompt | Trouble accessing the website, but not concerning anything dealing with WiFi |
Speaker | Customer |
Priority | High |
Other ways to use Tags
After your Configured and Generated tags have been processed for your conversations and you can see them being applied, you can use them to gain more insights in these conversations!
Filters - Configured or Generated Tags can be used as filters/saved searches, which can be used to set up grading assignments.
Analytics - The Analytics tab can track the top tags over a time period. It will also list the percentages of conversation that each tag takes up.
