Open-ended question

An open-ended question is a survey question without answer options: the respondent types the answer in their own words, as in "What would you improve in our service?". It is used to collect reasons, ideas and experiences that a fixed list would miss. You gain depth, but you also add work when it comes to analyzing the answers. You need to code them - turn text into categories. When to use such a question and how not to overdo it, below.

Closed questions give you percentages right away. Open-ended ones give you raw material. They are often mixed: first "Rate from 1 to 5", then "Why?" - an open field. You get both a number and a reason. On the difference between the types, see the article on open vs. closed questions.

What an open-ended question is

Open-ended question - a question without predefined answer options. The respondent writes in text: a single word, a phrase, or a whole paragraph. It is used for opinions, reasons, suggestions, descriptions of experience. The analysis is qualitative: reading, grouping by meaning, coding into categories. Without coding, 500 answers stay a pile of text.

Roughly: "write what you think" instead of "pick from a list".

When to use it: ideas, reasons, pilot

Ideas and improvements. "What did you dislike?", "How would you improve it?" - a closed list cuts answers short. An open-ended question gives freedom. Handy for collecting Voice of Customer and feedback.

Clarification after a closed question. "Why did you choose this particular option?" - context for the numbers. It helps you understand what's behind the percentages.

Pilot. Don't know which options to offer? Start open - you'll collect typical answers, then turn them into closed options for the main survey.

Quotes and natural language. People write in their own words. For reports and for understanding the tone of your audience - exactly what you need.

For mass statistics without coding, open-ended questions are poor: 500 texts won't turn into percentages on their own. If you need numbers, use closed questions or rating scales.

Pros and cons

Pros: rich information, unexpected answers, natural language, you can catch things you hadn't thought of. You don't impose a framework - the person decides for themselves what to write.

Cons: a burden on the respondent (they have to type), a lower response rate, more skips. Analysis is labor-intensive - reading, grouping, coding. There's also the risk of social desirability - people write what's "right" rather than what's honest.

The sweet spot is 1-3 open-ended questions per survey. At the end of blocks or after key closed questions. More than that and people get tired and drop off.

Wording: more specific and shorter

A vague question gives a vague answer. "Your wishes?" - unclear about what. "What would you improve about delivery?" - now there's a focus. "Describe the last time when..." - better than "Tell us about your experience" in general.

Limiting the length helps. "In 1-2 sentences" or "up to 500 characters" - easier for the respondent and easier to analyze. Without a limit you get an essay or "all good".

Leading wording is bad. "What did you like?" - fine. "What did you especially like about our new design?" - already nudging. More on this - leading question.

In SurveyNinja: a text field and export

An open-ended question in the SurveyNinja builder is the text field question. You add it from the elements panel and can make it required or not - open-ended questions are often left optional. It's all in the element settings.

Answers go into reports and responses - you can see the text of each respondent. Export to Excel/CSV - a separate column per question, rows are respondents. From there it's convenient to code: sort, group, assign categories. On the Basic plan and above you can apply filters before exporting - for example, completed responses only - so you don't pull in extra data.

Coding: from text to categories

You read the raw answers and group them by meaning. Coding - you assign a category to each one (for example "complaint about delivery", "praise for the assortment"). After that you can compute shares, build cross-tabulations, segment. How to do it - in data coding.

Coding is usually done in Excel or dedicated software. Sometimes sentiment analysis helps - a first automatic pass. Two coders are more reliable: they cross-check categories and resolve disputed cases.

The NPS + "Why?" combo

A typical setup: NPS 0-10 (closed), followed by "What mainly made you give that score?" (open-ended). You get both a number - the NPS score and the shares of promoters/passives/detractors - and context: why the low ones, why the high ones. The open field is better left optional, otherwise those who don't want to type will skip it or leave.

Common mistakes

A pile of open-ended questions in a row. Three or four text fields - overload. Alternate them with closed questions and scales.

Vague questions. "Comments" - about what? "What would you improve about delivery?" - now that's clearer.

Skipped the coding. 300 answers with no grouping are garbage. Set aside time for the analysis, or run a pilot to narrow the range.

Example: a post-purchase survey

A survey after an order: closed questions "Rate delivery 1-5", "Did you like the product?" - and an open-ended "What would you change?" at the end. 200 answers. Half are "all great" or empty. The rest are complaints about packaging, slow delivery, quality. You code them into 5-6 categories and compute the shares. Plus a couple of live quotes for the report. That's how an open-ended question complements closed ones rather than replacing them.

If you only used closed questions, you'd get percentages on a scale, but you wouldn't learn about the cracked box or the courier being an hour late. An open-ended question catches what didn't fit into your list of options.