Text Analysis guide
Text analysis is the ability to leverage AI to analyze automatically all text/unstructured data in your Feedier account.
The module offer different features out of the box, such as:
Automatically generating summaries
Automatically detecting new topics
Tracking and analyzing topics (with sentiment analysis and link to attributes)
Follow topic trends
Leverage Feedier’s filtering technology as in the Dashboards, Reports, Segments, etc.
How to set up Text Analysis?
Text analysis must be enabled from an administrator user in the “Roles & Capabilities“ page.
Data is analyzed by the Text Analysis module
The module is automatically using all text/unstructured data present in the teams the user has access too. So, as long as the account has open text questions, Text analysis will bring value and show automatically new topics and measure sentiment.
Sentiment analysis in Feedier
The sentiment score is a quantitative metric that is any number between 0 and 100, where 0 typically indicates extremely negative sentiment, 100 represents extremely positive sentiment, and values in between reflect varying degrees of neutrality or mixed emotions. This score is widely used to evaluate the emotional tone in textual data.
Sentiment | Explanation |
|---|---|
61 to 100 | The verbatim is linked to a positive emotion on the part of the customer: Joy, Confidence, Serenity, Admiration. |
41 to 60 | The verbatim is not directly linked to an emotion or the emotion is not expressed strongly enough to be categorised. |
1 to 40 | The verbatim is linked to a negative emotion on the part of the customer: Anger, Contempt, Sadness, Disgust. |
For the sake of the simplicity, the score is displayed with a label:
Positive From 61 to 100
NEUTRAL From 41 to 60
Negative From 1 to 40
Topics in Feedier
A topic in the Feedier Platform is a way to group different answers/text based on a common meaning.
Topics in the Feedier Platform can be:
Automatically suggested with AI or created manually from a Feedier user
Tracked over time
Organized in sub topics
Summarized
Explored through its answers, linked topics and linked attributes
It’s important to note that:
Topics can be given specific instructions based on your needs in the Edit section
Topics are by default automatically attached by Feedier, but they can be detached or attached manually by a User.
How is the sentiment score calculated for a topic?
This is the total average of sentiment scores of verbatim related to a topic.
Formula:
Sentiment score ranges:
Positive → 61 to 100
Neutral → 40 to 60
Negative → 1 to 39
Examples:
| Positive Answers | Neutral Answers | Negative Answers | Sentiment Score |
|---|---|---|---|---|
Topic 1: Improvements | 30% * 97.5 (avg) | 40% * 50 (avg) | 30% * 2.5 (avg) | 50 → Neutral |
Topic 2: Facilities | 71% * 90 (avg) | 18% * 50 (avg) | 11% * 10 (avg) | 74 → Positive |
Topic 3: Management | 20% * 75 (avg) | 0% * 50 (avg) | 80% * 5 (avg) | 19 → Negative |
Topic Precision Score
The Topic Precision Score shows how accurately our AI is assigning responses to a topic.
It helps you:
Understand how well the AI is performing in terms of matching
Identify when responses are matched incorrectly
Improve accuracy over time by manually attaching or detaching responses
The accuracy percentage updates automatically as you make changes, giving a clear signal of AI reliability for each topic. When a text answer is attached or detached manually from a topic the topic precision score will update.
Accuracy (%) = (Linked Responses to a topic − Manual attach/detach) ÷ Responses linked to the topic × 100
From Text Analysis to actionable Insights
From global to specific
The main important feature of the Text Analysis module is the ability to filter in real-time all the insights based on your specific context (time frame, attributes, source, team, etc.).
Generated topics and summaries will be more valuable when narrowing down the scope with the Feedier filter technology.
Detecting pain points
Two elements are extremely important when it comes to pain points in Text Analysis:
The painpoint itself
Its impact on the business
Feedier helps you do to both in one place.
Identifying pain points
There are two different methods:
Use global topic names (such as product, quality, service, sales, etc.) and add a sentiment score filter (positive or negative) to only match outliers in the Text analysis module.
Use specific topic names (such as product bugs, product ideas, service issues, sales bottlenecks) to only match outliers.
In both cases, Feedier will be able to identify clearly the pain points for you.
The impact of every single topic can be found on the left panel with its volume over time:
Indentifying root cause
For every topic, Feedier provides a comprehensive view with information such as:
Feedback
Related attributes
Related topics
When it comes to root cause, the attributes related to the topics are the most actionable insights. It means that for every topic (based on the filters set), you can easily identifty common similarities.
Creating Action Plans
Action plans with key improvement and pain points can be generated for every single topic by Feedier.
Leveraging Feedier reports
Insights coming from topics and text answers can be integrated with the Feedier Report module.
It’s possible to easily display:
Top X topics or answers based on sentiment
Top X topics or answers based on volume
Last but not least, every widget in the report can be filtered based on time, source, attribute or any other filter.