In Conversation Intelligence, Tags saved searches, and streams work together to provide customers with the deepest level of insight about their business. It is important to configure and set up each product as you implement Conversation Intelligence. Here are high-level definitions:
Tags are used to track if something happens in a conversation. It is always a yes/no question. It gives you access to Analytics, where you can track how often something is happening in a conversation.
Saved Searches are used to quickly access high-priority conversations that meet specific criteria. It is essentially altering without proactive alerts.
Streams are used to understand the "why" and "how," open-ended questions, not the yes/no. It can be based on tags but is leveraged to understand why something is happening in a conversation.
__________________________________________________________________________________
Tags let you identify whether or not something is occurring in a conversation. It is a word or phrase that occurs in a conversation. They are one of the foundational criteria you will use to break down your customer conversations.
What should be set up or configured as a tag?
Any word or phrase that you want to be able to easily filter on in Pathlight
Any word or phrase that you want to be able to count the occurrence of in a chart based on a specified timeframe using Analytics
Tags are the basis for being able to run saved searches
Here are examples of tags:
“Cancel”
“Refund”
“Technical Issue”
“Too expensive”
Here are examples that are NOT tags:
“General Packaging Feedback”
“Shipping process”
“Sign up flow”
__________________________________________________________________________________
Saved Searches is a feature where you can save specific filtered criteria that are important for you to monitor on an ongoing basis. Here you are able to track when certain items, like tags, are happening in conversations. It is used as a shortcut to bring the most important information front and center to your entire team in one click.
Note: Saved Search often includes multiple tags, as well as other criteria (negative customer sentiment) to bring the most critical conversations front and center.
What should be set up as a saved search?
Any critical grouping of conversations that you want to be flagged without having to spend time applying filter criteria each time you are in CI
Saved searches will be eventually used for altering, so think about what types of conversations you would want to be altered on, and use that to determine what becomes a saved search.
For example:Maybe you care about and want to be notified immediately when a customer writes in about a delivery issue.
In this scenario, you would create a saved search for any tag/tags that have to do with a negative reaction, giving your team the ability in one click to pull up ALL conversations that make the saved search criteria so you can review those most important tickets.
You can also export this grouping of conversations.
__________________________________________________________________________________
Streams programmatically extract top-level themes and trends from customer conversations, organizing them based on content using AI. This tool provides insights into the “content” of your customer conversations. Streams are specifically designed to answer open-ended "why" and "how" questions, offering a nuanced understanding beyond simple yes/no queries.
Streams are generative and require no manual configuration beyond an initial general description of the focus area, like "Why are my customers churning?"
They automatically categorize conversations into themes and sub-themes, revealing insights and covering blind spots in your understanding.
Additionally, Streams allow for filtering and categorization based on the themes and sub-themes they produce, facilitating a deeper analysis of customer interactions.
What should be set up as a custom stream?
Custom Streams should be established based on high-level topics or categories, which are not necessarily specific words or phrases from customer conversations.
They are useful for explaining "why" certain tags are trending.
Example: If you notice a 20% increase in the "cancel" tag, you can set up a Stream to delve into the primary reasons behind these cancellations.
Here are examples of Streams:
“General Packaging Feedback”
“Shipping process”
“Sign up flow”
__________________________________________________________________________________
Intents are designed to pinpoint the primary reason a customer conversation takes place.
Examples of such customer intents include “Account Administration,” “Shipping Issues,” and similar specific purposes of customer interactions.
Intents can be identified through one of three methods:
Utilizing the default stream specifically for analyzing customer intents, which is the recommended approach.
Defining a set of predefined intents, similar to configure tags.
Permitting the AI model to autonomously and randomly determine one.
With the recommended approach of using Streams to detect customer intents is the recommended approach, as it requires no upfront configuration work. This method allows the identification of conversation intents to adapt automatically to any new reasons customers may begin to require support.
Use case | Feature |
I want to identify conversations with high-priority issues that we already know about - such as churn risk, cancellations, and refunds being issued. | Tags - use tags to filter to a known issue through any cut of time |
I want to keep track of how often something is happening | Tags - since you already know what to look for, review tags in analytics to see how they are trending |
I want to know why customers are contacting us | Streams - tell you what you don’t know yet by generating themes and sub-themes from your conversations |
I want to understand the reasons that drive a high-priority issue like refunds. We are getting refunds, but why are we getting them? | Streams - with ‘reflections’ (essentially a sub-theme), you’ll see all the underlying reasons for a particular issue |
I need an answer to a question like - “Why are customers canceling their subscriptions?” | Streams - reflections will answer these questions, similarly to the above example |
I want to group together a combination of tags and view conversations that relate to a high-level issue like ‘Churn’ | Saved Searches |
I want to be alerted when something happens | Saved searches - you can tie an alert to a saved search. Any time a new conversation matches that search, we’ll send an alert to Slack |
__________________________________________________________________________________
We hope this was helpful! Please submit a ticket here if you have any questions or need further assistance.