Journey Insights Hub uses natural language understanding to identify the meaning, themes, and relationships within open-ended feedback.
Instead of relying only on keywords or requiring teams to manually define every topic in advance, the tool learns directly from the language in your data. This makes it possible to analyze sources such as survey responses, reviews, support tickets, chat transcripts and employee feedback quickly and at scale.
Learn the language of your data
When you create a text analysis project, Journey Insights Hub analyzes how words and phrases are used throughout the dataset. It learns:
Which concepts are related
How terminology is used in context
Which topics are most relevant to the feedback being analyzed
For example, customers may describe the same issue using phrases such as “slow response,” “long wait,” “delayed reply,” or “it took days to hear back.”
Rather than treating each phrase as unrelated, Journey Insights Hub may recognize that they all refer to response time.
Why this matters: Journey Insights Hub can organize feedback according to meaning, even when different people describe the same experience in different ways.
Understand words through context
Journey Insights Hub represents words and concepts using mathematical models called word embeddings. Each embedding contains hundreds of dimensions that capture different aspects of meaning and nuance.
Words and concepts used in similar ways are positioned closer together within a semantic space. This semantic space represents the concepts found in the project and the relationships among them.
For example, the word “support” could refer to:
Assistance from a service team
Technical compatibility for a feature
Internal support from a manager
Journey Insights Hub evaluates the surrounding language and the broader context of the project to understand how the term is being used in a particular response.
Similarly, the word “issue” might refer to a billing error, login problem, product defect, or delivery delay. Context helps the platform distinguish among these experiences and reduces the limitations of keyword searches, which may overlook related wording or combine unrelated uses of the same term.
Combine general language knowledge with domain-specific terminology
Journey Insights Hub begins with an existing understanding of language. Its natural language modeling technology draws on a background semantic model containing general relationships among words and concepts.
The background space contains general relationships between concepts and reflects common knowledge about how words and ideas relate to one another. It then applies that general understanding to your text analysis project data. Through a process known as transfer learning, it adapts its existing language knowledge to the terminology used in your organization, industry, or research area.
For example, customers may refer to a feature using:
Its official name
An abbreviation
An internal product label
An informal description
By examining how those terms appear throughout the feedback, Journey Insights Hub can begin identifying that they refer to the same or closely related parts of the experience. It can also begin recognizing emerging terminology, such as the name of a newly released feature or a phrase customers have started using to describe a recent problem.
Since the platform combines existing language knowledge with the patterns found in the project, it does not require supervised training or thousands of manually labeled examples to begin producing useful analysis.
How this differs from traditional ontologies
An ontology is a structured collection of terms, categories and relationships created for a particular subject area.
Ontologies can provide detailed and accurate domain knowledge, but they often require specialists to create, maintain, and update them.
For example, a customer-service ontology may include predefined categories such as billing, technical support, account access, and delivery. When customers begin discussing a new feature, policy, or recurring problem, those terms and relationships may need to be added manually before the system can classify them correctly.
Journey Insights Hub takes a different approach:
Traditional ontologies | Journey Insights Hub |
Use predefined terms, categories, and relationships | Learns terminology and relationships from the data |
Often require specialists to create and maintain them | Reduces the need for manual setup |
New terminology may need to be added manually | Can begin recognizing new language through context |
May require labeled examples or extensive preparation | Allows teams to begin analyzing feedback more quickly |
Apply an established structure to the text | Builds a model reflecting the language used in the project |
As a result, teams can begin analyzing feedback without first building a large custom ontology or labeling thousands of examples. Concept lists, Science assertions, and other configuration tools can then be used to refine the analysis around established business definitions and project requirements.
From language to insights
Once Journey Insights Hub creates a semantic space for a text analysis project, Journey Insights Hub uses that model to help you explore and quantify your feedback.
You can use the platform to:
Identify prominent themes and concepts
Discover relationships between topics
Understand sentiment toward different aspects of an experience
Compare feedback across segments or time periods
Find issues that are emerging within the data
Explore the records behind individual findings
For example, an analysis may reveal that negative feedback about account access is closely connected to password resets, verification steps, and delayed support responses.
It may also show that these concerns appear more frequently among new customers or increased after a recent process change. The result is a structured view of qualitative feedback that helps teams move from large volumes of text to findings they can investigate and act on.