Records connects the visualizations to the underlying text. Use it whenever a concept, relationship, score association, or sentiment result requires closer review.
The visualizations help identify patterns. Records helps you determine whether the pattern is consistent, understand the language people used and recognize exceptions or context that a chart cannot show.
Open matching records
The Records panel reflects the concepts and filters currently applied to the text analysis project.
Each record card provides a preview that may include:
A record title or identifier
Selected metadata
An excerpt of the text
A pin control for keeping a useful record readily available
Select a card to open the full record viewer.
View exact and conceptual matches
When a concept is selected, use the match-type controls in Records to review its exact and conceptual matches separately.
Exact matches contain the selected word or phrase, including basic variations in spelling, capitalization, or word form.
Conceptual matches contain language that Journey Insights Hub identifies as meaningfully related to the selected concept, even when the exact wording does not appear.
Start with exact matches when you need to confirm how people explicitly used a term. Then review conceptual matches to understand the broader language surrounding the same idea.
Conceptual matches are based on the meanings and relationships learned from the text analysis project. Review several examples to confirm that the related language represents the intended topic rather than a different use of the concept.
The active filters remain relevant when switching between match types. This allows you to compare exact and broader discussion within the same group of records. If conceptual matches repeatedly introduce an unwanted meaning, consider refining the concept or creating an advanced concept that excludes the unrelated language.
After confirming the active concepts, match type, and filters, you can:
Select Download matching records to export the current subset for offline review, reporting, or additional analysis.
Select Create project to create a focused branch project containing the matching records.
Download the records when you want to preserve or share the evidence without creating another text analysis project. Create a branch project when the subset requires deeper, ongoing analysis and you want Journey Insights Hub to model its language independently.
Read the full record and metadata
The record viewer displays the complete text alongside its associated metadata.
Use the navigation controls to move through the matching records without closing the viewer. Compare several records rather than using a single quotation to explain an entire pattern.
Review the metadata to understand who or what the record represents, when it was collected and which category or score it belongs to. This context can explain why similar language has different meanings across records.
Choose which metadata fields to display
Select Select fields in Records to control which metadata appears in the side panel.
Choose fields that help you interpret the records, such as:
Date
Product or service
Region or channel
Score or rating
Customer or employee segment
Case type or source
Keeping the most relevant metadata visible makes it easier to compare records without crowding the panel with fields that do not support the current question.
Validate a finding
When reviewing a result:
Open records from the relevant concept or visualization
Read examples that support the apparent pattern
Look for exceptions or alternative meanings
Compare records from different metadata groups
Confirm whether exact and conceptual matches reflect the intended topic
Check whether the result is based on enough records
Save or export representative evidence only after reviewing the full context
This is especially important for sentiment, conceptual matching, filtered subsets and concepts with a large score difference but few matches.
Create a focused text analysis project from matching records
When a filter is active, a create project option appears in Records beside the download option.
Select Create project to build a new Journey Insights Hub text analysis project containing only the records that match the current concept and filter selections.
This is useful when you want to:
Investigate one segment in greater depth
Separate a focused subject from a broad master text analysis project
Build a text analysis project around a specific product, region, audience, score range, time period, or concept
Reduce background language that is unrelated to the new question
Allow Journey Insights Hub to model the language of the selected subset as its own text analysis project
Because every text analysis project is modeled from its own records, the new one may surface concepts, prevalence, relationships and analytical patterns that were less visible in the broader project.
Before creating the text analysis project
Confirm:
The intended concept selections are active
Every filter condition is correct
The remaining record count is large enough to support analysis
The selected subset represents the question you want to investigate
You have not accidentally carried over an unrelated filter
The new text analysis project will have a clear, descriptive name
The new text analysis project contains only the matching records. It does not preserve the excluded records as a comparison group.
If comparing the subset with the rest of the data is important, complete that comparison in the original text analysis project before branching. Unique to this filter can help identify concepts that distinguish the subset while the wider context is still available.
Filter or create a new text analysis project?
Continue using a filter when:
You want to move quickly between groups
You need to compare the subset with the rest of the text analysis project
The question is temporary or exploratory
You want to keep one shared set of concepts and relationships
Create a new text analysis project when:
The subset has become a separate analysis
You want Journey Insights Hub to model that group independently
The broader text analysis project contains too much unrelated language
The focused text analysis project will be used repeatedly or by a different audience
Creating a focused text analysis project should support a clear analytical purpose rather than replace ordinary filtering for every comparison.
Download matching records
Select Download matching records to export the records that match the current selection.
Choose either:
XLSX
CSV
The download includes up to 20,000 matching records. Use the API when you need to retrieve more than 20,000.
Before downloading, confirm the active concepts and filters. The exported file reflects the current matching set, so an overlooked filter may exclude records you intended to include. Use a descriptive filename that identifies the text analysis project, concept or segment and date range. This makes the exported evidence easier to trace back to the analysis.
Use Records throughout the workflow
Records should not be treated as only the final step. Return to the source records whenever:
A conceptual match seems unexpected
A sentiment result appears inconsistent
A Drivers result is based on relatively few matches
Two filtered groups show a large difference
A relationship in Galaxy needs explanation
You need representative quotations for reporting
You are deciding whether a filtered subset deserves its own text analysis project
Reliable insights come from combining the scale and structure shown in the visualizations with the detail and context available in the original records.