With powerful AI tools like Text Analysis, it’s easier than ever to analyse open-text responses. However, a question that often pops up is how to phrase survey questions in order to maximise the value from such tools. In this article, we’ll look at all the features available to you with Text Analysis, and give you open-ended question examples to get the most out of them.
Open ended vs closed ended questions: Unlike open-ended questions, closed-ended questions give people a predefined set of answers to choose from. Open ended questions, on the other hand, let people reply to a question with free text. This allows survey respondents express opinions, thoughts and feelings in a way that closed questions don’t. This makes them incredibly valuable for sourcing qualitative insights that help you to understand exactly how you can improve your business.
Why use keyword extraction?
Keyword extraction is powerful technology that extracts key words from an answer. It helps to reduce complexity when analysing open text and give you a quick overview of what respondents are talking about. It’s also great for spotting trends across answers.
A general rule for open ended questions, that also applies here, is not to try and steer respondents in any direction. You should also avoid questions where a respondent can answer with ‘Yes’ or ‘No, or just a few words. These are not suitable for keyword extraction or Text Analysis in general.
Good open ended question examples
- ✅ Tell us about your last experience with us
- ✅ Elaborate more on why you gave this rating
- ✅ What can we do to improve our product?
- ✅ What are the top-5 reasons for using our services?
These are examples of very open questions that let the respondent write whatever comes to mind without biasing their answer.
Why use sentiment analysis?
Sentiment analysis is useful for quickly understanding how people feel about your company or product. Are their feelings positive or negative? Do they like your offering? Sentiment analysis can read text answers for you and automatically assign a sentiment label, either ‘positive’, ‘negative’, ‘neutral’, or ‘mixed’. In order for sentiment analysis to be most effective, you need to phrase your questions in a neutral way that doesn’t push respondents to answer in a positive or negative way.
Bad open ended question examples
- ⛔️ What was the best part of your experience
- ⛔️ What are the top-5 reasons for using our services?
These questions are not suitable for sentiment analysis, since there is already a positive bias in the question. This will affect the way the respondent answers, potentially skewing the sentiment.
- ❓“Elaborate more on why you gave this rating” (for (e)NPS question)
The question is not bad in itself, but given that it is a follow-up question to a NPS question, you might as well use the NPS score for detractors/passives/promoters to determine the sentiment.
Good open ended question example
- ✅ How was your latest experience with us?
- ✅ What did you think about the service you received?
These are questions are good because they are very open in terms of the sentiment they solicit. The only risk is that the respondent writes an excessively long text (e.g. X was good, Y was bad, Z was terrible), potentially giving mixed sentiments in the answer. If possible, it’s therefore beneficial to either limit the answer text length or limiting the question like:
- ✅ In one sentence, describe your latest experience with us, focusing on the most distinct aspect that influenced your impression
Why use topics?
Topics allows you to automatically sort free text answers into a set of topics you define yourself. Combining topics with other data such as sentiment gives more insight to what the respondents are talking about and how they are feeling.
Bad question examples
- ⛔️ What are the top 5 reasons for using our services?
This questions encourages the user to enumerate the answers, giving a high chance that the text answer will consist of 5 separate topics. Since a text answer only can be mapped to one Topic, the topic assigned may seem arbitrarily chosen.
Good question example
- ✅ In one sentence, describe your latest experience with us, focusing on the most distinct aspect that influenced your impression.
This is a good example of an open question that reduces the risk for ambiguity by limiting the answer to one sentence.
The way that a question is formulated may limit the value that the different Text Analysis features will give. As seen in the examples above, a question that is good for one feature might be bad for another. Rather than struggling to create a universal question that works for all features, it’s likely easier to adjust your questions to the elements relevant to the question type.
If you’re interested in learning more about Text Analysis, you can download our quick start guide here. Alternatively, book a demo of Netigate and a friendly member of our team will give you a live demonstration of how it all works.