This article will provide you with a comprehensive guide on how to analyse open-ended (OE) responses on the platform.
- How to analyse OE responses?
- How to analyse OE responses on the platform?
- How to use the Automated Open Ended Tagging tool on the platform?
- Some limitations on the word cloud and word count
This article will provide you with a comprehensive guide on how to analyze open-ended (OE) responses on the platform.
How to analyse open-ended (OE) responses?
To ease the analysis and report preparation, usually we will group the open-ended responses into buckets or categories. Here are a few ways on how to group responses (AKA OE coding):
- Grouping the answers as it is
Example question: When thinking about fast food, what brands come to your mind?
The best example for this scenario would be a top of mind brand question. The respondents might have some typos or put in the abbreviations instead of the full brands. Further grouping or organizing the responses would ease the data analysis. - Grouping into product categories
Example question: What is the brand that you think represents your values the most?
Since we didn't specify or limit to a brand category, the respondents are free to input any brands they like. In this case, it might be not so useful to group the brands as it is because the count may be too low. We can group the brands by their product category along with some top brands mentioned instead, e.g. clothing brands (e.g. Uniqlo, Zara etc.), sports apparel brands (e.g. Nike, Adidas etc.). - Grouping into broader topics / subjects
Example questions: What do you like about brand X? / If you were to use a phrase to describe brand X, what would that be?
Every consumer cares about different aspects, hence will most likely have a different perspective for a brand. In this scenario, we can group their responses by some broader categories, e.g. price related, product features related, brand value related etc. In our topline analysis, we can determine what is our strength / what people like about us easily. If necessary, we can then further look into their verbatim and understand what the consumers mentioned. - Grouping by sentiments (positive, neutral, negative)
Example questions: What do you think about the initiatives carried out by brand X?
The responses we will get from this question are likely their thoughts and might be difficult to group into broader topics / categories. Grouping into sentiments might be more suitable for this type of questions, to help us gauge people's opinions.
Grouping the answers into categories
When the answers are too scattered to group by itself, we can consider grouping them into the respective categories.
How to analyse OE responses on the platform?
In the dashboard, you will have 4 modes of viewing the open-ended responses.
- Word Cloud
Compose of words mentioned in the respondents' answers, in which the size of each word indicates its frequency (*limitations)
- Word Count
Show the words mentioned in the respondents' answers, along with the frequency (*limitations)
- Full Responses
Verbatim responses that were written in the survey
- Tags
Allow users to automate grouping / tagging of the responses directly into buckets. The data can then be shown as tags using bar chart or word cloud mode. Click on "Tagging" to start assigning tags to the responses!
How to use the Automated Open Ended Tagging tool on the platform?
- By clicking “Tagging” you will be brought to the tagging page where a short click through tutorial will be displayed.
- When ready, click on “Regenerate tags”. The first time tags are generated, you will only be able to select “Generate New Set of AI Tags”. You will then see the blue icon “Generating tagging” pop up. The automated open ended tagging is now in progress.
- Once tagging is complete, you will be able to see all the AI-generated tags below. You can also take this opportunity to preview the data either in bar chart or word cloud mode.
- If you are happy with the results, you can click “Back to Insights” which will bring you back to the insights dashboard. Else “Back to Tagging” will return you to the tagging page.
- You can also view all full responses that are classified under certain tags.
- If a response is not assigned a tag, it will be tagged as “No tags”. In this case, you can either assign it to an existing tag, or assign a new tag to it via the “Enter text” box.
- The automated tagging feature supports multiple languages. Tags will appear in the dominant language used in the responses. So, if most verbatims are in Malay, tags will be in Malay. If more are in English, tags will show in English. It follows the majority language—it doesn’t translate. However, if you would like to use a specific language, feel free to rename the tags in your language of choice.
- Below is an example of automated open ended tagging where responses of different languages are given tags based on their sentiment or overall theme.
Some limitations on the word cloud and word count:
- The word cloud and word count analyses word per word. In a top-of-mind brand question, the brand name would be split to multiple words under the following circumstances:
- If the brand name consists of 2 words or more, e.g. Cotton On, Burger King, Pizza Hut etc.
- If the brand name was mistakenly spelt as 2 words or more, e.g. Kit Kat (instead of KitKat), 7 Eleven / Seven Eleven (instead of 7-Eleven) etc.
- If the brand name was spelt incorrectly or differently, e.g. McDonald, McD, McDonalds, McDonald's. All these would not be treated as separate words, i.e. these words would appear separately in the word cloud and word count.
- The word cloud and word count are not available for the languages that are not using Roman/Latin scripts (e.g. Thai, Chinese).
- The word cloud and word count are not optimized for non-English responses, so you may see some 'stop words' (i.e. unimportant words in a sentence) included. Some examples of stop words in Malay are 'beberapa', 'yang', 'ialah' and 'itu'.
Reach out to us using the chat feature below if you have any questions!