Menu → Text Analysis
In short: Text Analysis automatically classifies your verbatims by topic and sentiment. You create a taxonomy (your classification plan), fine-tune it in the Taxonomy Lab, then publish it.
Key concepts
A verbatim is a free-text response. Text Analysis relies on four nested concepts:
Concept | What it is | Example |
Taxonomy | Your classification plan: a set of segments. You can have several, for example one per journey or per business line. | "Passenger journey" |
Segment | A group of related topics, matching a stage or an area. | "Arrival and baggage" |
Topic | A specific subject that AI spots in verbatims, based on its name and rules. | "Delayed baggage" |
Insight | A recurring finding that Feedier surfaces within a topic. | "Customer service unreachable" |
Starting from scratch? Follow the guide Build your topic plan from the customer journey.
Who can use it?
Role | Access |
Admin | Full access: create, edit, publish and delete taxonomies, use the Taxonomy Lab. |
Editor | View the analysis, create and edit topics, correct a verbatim's classification. The Taxonomy Lab and publishing are reserved for admins. |
Viewer and Restricted Viewer | Read-only. |
Access to the page also depends on the Text Analysis page module, enabled per role in Settings → Access management.
Read the Text Analysis page
The page has three tabs:
Tab | What it's for |
Taxonomies | The overview, with one card per segment. |
Topics | The list of all topics, to manage them one by one. |
Verbatims | The list of verbatims, with their topics and sentiment. |
In the top right, three indicators summarize the analysis coverage for the selected period: Total verbatims, Untagged verbatims and Untagged rate.
A rising untagged rate signals that a new subject is appearing in your verbatims and no topic covers it yet.
In the Taxonomies tab, choose the taxonomy from the My taxonomies list. Each segment card shows:
the number of verbatims in the segment and their share of the total;
the dominant sentiment and the NPS;
its topics, each with its volume and its business impact score out of 10.
The Sort by button changes the order of segments. Apply a filter and the period selector narrow the analysis to part of your feedback.
Explore a topic
Click a topic to open its page: volume, sentiment and NPS over time, then four tabs.
Tab | Content |
Topics | The topic's trend for the KPI of your choice. |
Attributes | The verbatims' context, for example the customer type or the point of sale. |
Verbatims | The topic's verbatims. The passage about the topic is highlighted and has its own sentiment. |
Sources | The most represented sources. |
On the right, the Insights panel lists the findings that stand out, sorted by volume, with the All, Negative and Positive filters. Click an insight to add it as a filter.
Create a taxonomy
In the Taxonomies tab, click + next to My taxonomies.
Option | When to choose it |
Create with AI | In most cases. AI suggests a structure based on your verbatims and your instructions. |
Build manually | You already have your exact list of segments and topics. An empty draft opens in the Taxonomy Lab. |
Create with AI, step by step
Step 1, Context. Name the taxonomy, then choose the Analysis type: Customer journey (from first contact to loyalty), Business areas (your departments or functions) or Operational areas (your processes, such as billing or delivery).
If needed, specify the scope limits (for example "Only cover the post-purchase experience") and what you want to achieve. Both fields are optional. Click Next.
Step 2, Semantic requirements. Indicate whether you have topics to include: None, Plain text (you list your topics and AI structures them) or Import an existing taxonomy (JSON).
Open Fine-tune the size and structure of your taxonomy to set the number of segments and the total number of topics.
In Sources, leave All sources on or choose the sources to analyze.
In Visibility, choose the teams that will have access to the taxonomy.
Click Generate draft. It opens in the Taxonomy Lab.
A taxonomy's topics are only detected on feedback from its sources. If a taxonomy only concerns one survey, limit it to that source.
Fine-tune a taxonomy in the Taxonomy Lab
The Taxonomy Lab is the workshop where you create and edit a taxonomy. You work on a draft tested on a sample: the live version doesn't change until you publish.
To open it, select the taxonomy then click Taxonomy Lab.
Step | On which data | What you do |
1. Draft | A sample of 1,000 verbatims | You edit segments, topics and rules, and see the effect immediately. |
2. Validation | 10,000 verbatims | You check the result on a larger volume. |
3. Publishing | All data | The new version is applied to your verbatims and becomes the live version. |
Write a topic's rules
Click a topic to open its rules. They tell AI when to link a verbatim to the topic.
Rule | Effect | Example for "Ticket prices" |
MUST | The verbatim is tagged if it matches. | "Mentions the price or fees of a ticket" |
MUST NOT | The verbatim is never tagged if it matches. Only describe the case to exclude. | "Only talks about baggage prices" |
Below the rules, Examples from the last test shows sample verbatims linked to the topic, with the relevant passage highlighted. If an example doesn't belong, exclude it, adjust the rules, then click Save to draft.
Tip: a topic with an explicit name often needs no rules at all. Save rules for your internal vocabulary and for similar topics you need to tell apart.
Handle untagged verbatims
The Untagged column groups by theme the sample verbatims that don't match any topic. For each group, Create a topic in… adds a topic to the suggested segment. Click Refresh themes after your changes.
Add, remove, group
Action | Where |
Add a topic | Add a topic, at the bottom of each segment |
Add a segment | Add a segment, at the end of the list |
Remove a topic | Remove topic from draft, in the topic's panel |
Work on specific feedback | ⋮ menu → Advanced sample filtering (IDs, period, attributes) |
Go back to a previous version | ⋮ menu → Version history |
Start over from the live version | ⋮ menu → Discard draft |
The Ask Feedier panel can review the structure, explain why a verbatim is tagged, tighten rules and suggest new topics. Each change is submitted to you for approval.
Validate and publish
When you're happy with the draft, move to the Validation step, then to Publishing.
Publishing has a cost. It re-analyzes your existing verbatims and uses analysis budget. The amount is shown before confirmation and, if it's large, the request waits for an admin's approval. See AI analysis budget.
Limit a taxonomy to certain sources
Open the Taxonomy Lab of the taxonomy in question.
Click Sources at the top of the page, turn off All sources and check the sources you want.
Apply: a new sample is drawn from these sources. The button then shows "N sources".
Validate, then publish. The new scope applies on publishing.
Choose which teams see a taxonomy
Select the taxonomy, then click the gear icon to open Edit taxonomy.
Open the Visibility tab.
Choose Selected teams (your team and the checked sub-teams) or Entire workspace.
Click Save.
Parent teams always have access to taxonomies created in their sub-teams.
Rename, reorder or delete a taxonomy
Action | How | Good to know |
Rename | Gear icon → pencil next to the name | Nothing else changes. |
Reorder | Drag the taxonomy in the My taxonomies list | The first one opens by default, for the whole workspace. |
Delete | Gear icon → trash can | Segments are deleted, topics are kept. |
The Details tab in the same panel shows the taxonomy's type, goal and constraints, as well as the verbatims analyzed and tagged.
Manage topics one by one
The Topics tab lists all your topics with their segment, taxonomy, number of verbatims, whether they have rules (Instructions) and their Precision score. Use the search, Sort by and Filters to find a topic.
Each row's ⋮ menu offers Open detailed view, Edit and Delete.
Correct a verbatim's classification
In the Verbatims tab, each row shows the verbatim's topics. Click + to add a topic, or the cross on a topic to remove it.
Limitations and behaviors to know
A topic's rules can be edited only in the Taxonomy Lab.
As long as a draft isn't published, the live version stays unchanged.
Each publication uses analysis budget.
A taxonomy's topics are only detected on feedback from its sources.
There's no team-specific default taxonomy.
The language of new topics follows the team language, set in Settings → Teams → Advanced. Changing the language doesn't translate existing topics.
Best practice: give topics neutral names, with no sentiment. "Delivery" is enough to analyze both positive and negative reviews.
FAQ
How do I set a team's default taxonomy?
This setting doesn't exist per team. The taxonomy placed first in My taxonomies opens first for the whole workspace. To make a team see only its own taxonomies, set their visibility.
Why is a taxonomy missing or empty in another team?
Check its visibility (the team must be selected) and its sources (its topics are only detected on feedback from those sources).
Does deleting a taxonomy delete its topics?
No. Only its segments are deleted. Topics are deleted separately, from the Topics tab.
How do I restrict a taxonomy to certain sources without re-analyzing everything?
Change its sources in the Taxonomy Lab, then publish. Feedback from removed sources keeps the topics already assigned: only new feedback is affected.
Why are my topics in a different language from the one my team works in?
New topics are created in the team language. Check it in Settings → Teams → Advanced before generating a taxonomy.
Why do some verbatims have no topic?
They're too short or too vague, or no topic matches them. The Untagged column in the Taxonomy Lab helps you spot missing topics.
Can I export a taxonomy?
Not the taxonomy itself. However, you can export your verbatims with their topics from the Feedback feed.
Last reviewed: November 11, 2026, version 4.23.5.


