📌 Text AI is available on the Professional or Enterprise plan.
What Is Topic Analysis?
Topic analysis is a feature where AI independently identifies and groups key opinions that recur within large amounts of text data.
For example, if the responses "The sound is clear" and "The audio quality is satisfying" are collected for the question "Why did you purchase this product?", the AI contextually understands these and automatically classifies them under a single topic called [Audio quality].
Why Should You Use Topic Analysis?
1️⃣ Quickly analyze large volumes of text data.
Manually classifying and organizing hundreds or thousands of open-ended responses takes a great deal of time. Using topic analysis, AI automatically analyzes responses and extracts common topics, dramatically reducing analysis time.
2️⃣ Derive objective, data-driven insights.
AI comprehensively analyzes all response data to generate common topics. This reduces variance caused by the analyst's subjective judgment and enables a more balanced understanding of overall data trends.
3️⃣ View analysis results intuitively.
Extracted topics are visualized in bubble chart form, letting you see at a glance how frequently each topic was mentioned. This makes it easy to quickly identify key opinions or critical issues and use them as reporting material.
When Do You Need Topic Analysis?
✅ When you want specific customer reactions after a new product launch
"Satisfaction scores are high, but I don't have the heart to read through 1,000 open-ended responses to find out exactly which points customers are satisfied with. I want to understand the 'context,' not just a few key keywords. 🧐"
🧙 Here's how you can use it
Running a new topic analysis has AI analyze responses based on the new data, automatically extracting the positive and negative topics that customers currently mention most. This lets you quickly grasp the key context of customer reactions without reviewing each open-ended response individually.
✅ When you want to track trend changes in a monthly brand survey
"Last month there were many complaints about 'price,' but I want to compare what feedback has increased after this month's response. Topic names change every time, which makes comparison difficult — is there a way around this? 🧐"
🧙 Here's how you can use it
Using the topics from a previous analysis allows you to compare data against the same topic standard. This makes it much clearer which opinions have increased or decreased over time, and you can also systematically track changes in customer perception of the brand.
How to Run Topic Analysis
📌 Topic analysis deducts 1 text credit per response.
1️⃣ Launch AI analysis 'Ossistant'
step 1. In the variable list on the left sidebar of the Text AI tab, select the text variable to analyze, then click the [AI Analysis] button in the upper right.
💡 Usage tip | Only variables that contain valid responses that have not been hidden can be selected from the variable list.
step 2. When the chatbot Ossistant appears, follow Ossistant's guidance and select 'Topic analysis.'
💡 Usage tip | 'Topic analysis' is only activated when text data is present.
2️⃣ Select the analysis method
Select either 'Run new topic analysis' or 'Use previously analyzed topics.'
Method | Description | When to use |
Run new topic analysis | Analyzes the characteristics of responses currently collected, regardless of past analysis history. | - When starting a new project or conducting exploratory research - Best when you want to independently understand current customer voices rather than compare with previous data. |
Use previously analyzed topics | Loads and uses topics generated from existing projects within the same Space. | - For regular/tracking surveys that use the same questions repeatedly - Essential when you want to ensure consistency in analysis standards and precisely track time-series trend changes. |
🧙 Understanding 'Use previously analyzed topics' through an example
Suppose you conduct a 'department store satisfaction survey' every month, and during the March analysis, the AI classified open-ended responses and generated the topic name 'Good accessibility.'
If you select 'Run new topic analysis' when analyzing data additionally collected in August, the AI may suggest slightly different topic names like 'Excellent accessibility' or 'Convenient transportation' based on the new response context.
Even if the meaning is similar, if the topic names differ, you may need to go through a separate matching process to directly compare the March and August data.
In this case, selecting 'Use previously analyzed topics' maintains the March-generated name 'Good accessibility' when classifying responses in the August analysis as well. This allows you to immediately visualize and accurately compare "how much accessibility satisfaction has changed since last March" without any additional work.
3️⃣ Start the analysis.
step 1. Before running the analysis, check the number of responses to be analyzed, the analysis feature to be used, and the credits to be deducted.
step 2. Select [Yes] to proceed. Selecting [Yes] starts the analysis.
Analysis generally completes within 2 hours; for 1,000 or more responses, it may take up to 48 hours.
Even if you close the browser window after requesting analysis, the analysis continues.
step 3. Once analysis is complete, you can view the key topics and opinions that commonly appear across all responses in the 'Topic analysis' area below the word cloud.
In the 'Survey data' area on the left, a 'Topic' label is added to variables selected for topic analysis, making it easy to see which variables have had topic analysis applied.
💡 Usage tip | Running topic analysis automatically creates a new variable in the left sidebar in the format [variable name selected for topic analysis]_[date topic analysis was run]. The analysis results for the generated variable can also be viewed in the Results tab, Cross-tab tab, and Report tab.
Using Analysis Results
1️⃣ Identify key topics with the bubble chart
This feature in the Text AI tab automatically classifies responses by topic and visualizes results in a bubble chart. The bubble chart can be downloaded as an image, and results can be viewed including the 'None' and 'Other' choices.
In the bubble chart, more frequently mentioned topics are displayed larger, and the number of responses mentioning each topic is provided. Clicking a specific topic navigates to that topic's response summary, where you can quickly and clearly find the information you need.
2️⃣ Understand context with response summaries
At the bottom of the chart, AI summarizes the entire responses grouped under each topic in 1–2 sentences. The summarized responses grouped under the same topic let you quickly grasp the key content.
Clicking the [∨] button next to each topic in the response summary shows up to 5 examples of original responses selected through the MMR algorithm.
3️⃣ Compare groups with cross-tab analysis
step 1. Click [Cross-tab] in the topic analysis results area.
step 2. In the cross-tab screen, the variable generated through topic analysis is automatically added as the analysis target. Adding the desired group or variable as analysis criteria here lets you compare and review topic analysis results by criteria.
Have you got a good understanding?
If you still have questions after reviewing this guide, please contact us anytime via the [Customer Support icon] in the bottom right corner of your screen. Our team will do its best to help you resolve any difficulties you're experiencing.