Analyze open text responses (with AI)

Open-text questions and map comments often give you the richest feedback in a survey, but with hundreds or thousands of comments, reading them all one by one isn't realistic. Maptionnaire has several tools to help, from AI tagging and AI summaries to search, word clouds and sentiment analysis. This article covers all of them and helps you choose which to use.

Which tool should I use?

• To see what people talk about, and how often: AI tagging. This is the best place to start when you have a lot of comments.
• To answer a specific question, such as "What are people saying about parking?": ask questions about your data with AI.
• To focus on one group or one place: use filters or spatial filters, then tag or ask the AI about only those responses.
• To find comments that mention a specific word: search.
• To read every comment, or copy them elsewhere: the response list in Charts.
• To get a quick visual of the most common words: word cloud.
• To get a quick sense of the tone: sentiment analysis.
• To share the results with others: the AI report.

1. Find themes with AI tagging

AI tagging turns a large set of open-text comments into a structured set of themes, without days of manual coding. It is the closest thing to what an analyst would produce by hand, just much faster. It works in two rounds:

1. It reads all the comments once and picks out the 5–10 most common themes.
2. It reads them again and tags each comment with the relevant themes.

How to tag responses with AI

1. Go to Analyze > Charts and open the open-text question you want to tag.

   

2. Click Generate Tags. All responses to the question are tagged automatically.

   

3. Review the tags and edit them if needed. The Tags tab shows all tags used across the responses to this question. You can also add tags manually.

   

Tags also show up in your Excel export, on their own "Tags" sheet, named {parent}_{question}_{tag}.

Why not just paste the comments into a general AI chat? General-purpose AI tools tend to cherry-pick and give results you can't reproduce. AI tagging goes through every comment in a structured way, and it is fully auditable: every tag, filter and prompt is visible, so anyone can retrace your steps and land on the same result. That makes it a good basis for a formal report.

Read more: Tag open responses with AI

2. Ask questions about your data with AI

When you want to know something specific, you can ask the AI directly: summarize the responses, list the top three concerns, or find out what people say about a particular topic. Then ask follow-up questions to dig deeper.

1. In the analysis tool, add the charts for the questions you want to analyze to your view. These can include open-text questions and map pop-up questions.
2. Open the Tools menu and choose Ask questions about data.
3. Check the table of responses. The AI only analyzes the responses visible in this table.
4. Type your question, or choose one of the default prompts, for example "Summarize the main reasons people like this area" or "Identify common suggestions for improvement".
5. Ask follow-up questions as needed.

What gets sent to the AI: the tool works on a "what you see is what you get" principle. Only the questions in your current view, with any filters you have applied, are sent to the AI. If you change your selection or filters, refresh the AI dialog so it works with your latest selection.

This is the most flexible of the AI tools, and good for exploring or for questions you didn't know to ask in advance. It is less structured than AI tagging, so use it for quick, specific questions rather than for producing a citable set of themes.

Read more: Use AI to summarize survey responses

3. Focus on a group or a place

Both AI tagging and AI questions become even more useful when you narrow down the responses first.

Compare groups with filters

Want to know if opinions differ by age, neighborhood or how people get around? Use the filter tool to select, say, "People who cycle to work", and then let the AI summarize the open comments only for that group. How to filter charts

Find out what people say about a specific location

If many respondents marked the same spot on the map, use a spatial filter to see what that's about:

1. In Charts, select the questions you want to look at, such as demographic questions and any pop-up questions linked to the map question.
2. Open Filters > Spatial Filters, draw a border around the area and click Apply.
3. All charts now show only responses from respondents who marked something in this area. You can see who they are (e.g. residents or small business owners) and what they wrote.
4. Open the AI tool and ask it questions, e.g. to summarize the comments related to this place.

4. Read, copy and search responses

View or copy responses as a list

Analyze > Charts shows the responses to open-text questions as a list. This includes comments made on map pins, lines and areas, which you can also view in Map responses.

Search for specific words

Open the question in Charts and type a word in the search bar to find the responses that mention it.

5. Visualize key words with a word cloud

The word cloud shows the 250 most frequently used words in the responses. You can exclude words (e.g. common words like "and") by typing them in the text field on the left.

6. Check the tone with sentiment analysis

Sentiment analysis classifies each response as positive, neutral or negative. It works best for questions where answers express an opinion, such as "What is your overall opinion on how this park is managed?". It needs no setup, so it's a quick first look at the tone, but it doesn't tell you why people feel the way they do. For that, use AI tagging or ask the AI.

The bar chart shows how the responses are split between positive, neutral and negative. Each response counts toward one overall sentiment.

The table gives a more detailed breakdown for each response. For example, "This place is important" might be 53% likely positive and 47% likely negative. The table includes:

• The overall sentiment (the one with the highest likelihood)
• The likelihood of each sentiment
• The original response
• The language the respondent used
• An English translation, for some languages

Note on languages: sentiment analysis supports a limited number of languages. Responses in other languages are automatically translated into English first, and the translation is shown in the "Translation" column.

Share your findings

• AI report: writes a full results report with a narrative, charts and map cluster analysis, which you can edit and share. How to
• Excel export: includes your AI and manual tags on a separate "Tags" sheet.
• AI-ready text export: exports your data with the survey structure as a .txt file for use in another AI assistant. How to

Check the results

AI tools save a lot of time, but treat their output as a draft. Check tags and summaries against the actual responses before you publish or share them, especially when a theme is based on only a few comments.

About Claude and data

As of August 2026, Maptionnaire uses Claude Sonnet 4.6, running on Amazon Bedrock — the same AWS infrastructure that hosts the rest of Maptionnaire. The model does not retain your responses after processing them, and does not use them to train models. All AI processing of response data happens in the same AWS region as your data — EU data stays in the EU, US data in the US.

For all AI tools in one place, see AI tools in Maptionnaire: overview.