Lifting the lid on Sentiment Analysis - what you need to know.

4 minute read  

We are seeing the rise of many highly technical tools to make sense of free text comments. Some sentiment engines position themselves in terms that are hard to get your head around – it all sounds like rocket science! So in his blog I would like to share what sentiment analysis is and how to use it in simple terms. 

Surveys are fundamentally either asking questions that are alpha numeric, I.E. choose a multi choice answer or a number such as 1 to 5. This is commonly termed as a Quantitative measure. Or alternatively they are asking for a free text comment in order to glean aspects of the respondents 'thoughts and feelings'.  This is known as a Qualitative data sample. 

Free text comments are known as 'verbatim'. Sometimes referred to as VOC which means Voice of the Customer. It's also worth noting that the feedback quality will be a direct result of the quality of your survey questions and survey structure. Remember 'garbage in, garbage out'.


Do Something with the Feedback

With technology enabling massive survey reach, thousands of comments and sometimes millions of comments can be captured - and that's great but it's useless unless the feedback can be utilised to improve, respond, recover or fix something. This is the key to all successful customer experience (CX) activity. You have to do something with the feedback you get!

Sentiment analysis provides a quick and effective way to sort all the comments into usable data. It uses machine learning or AI (artificial intelligence) to make sense of high volumes of feedback. 

Sentiment is about how we feel. Data is commonly sorted into positive and negative buckets. It can then be segmented into different areas such as staff service, products, complaints or for example, food type. 

For your organisation there will be specific areas of interest. These can easily be researched by the sentiment engine and the results from thousands of comments presented in drill down simplicity. 


Make the Sentiment Analysis usable

You can choose different ways to access the data, from word clouds to word counts. Word clouds can be single words (called unigrams), double words (bigrams) or triple words (trigrams). The benefits of short sentence analysis is that the context of a word’s use will provide insight. 

For example, in a positive bucket of feedback, 'food' and 'great' might stand out as common words. A bigram might show that 'great food' stands out and a trigram might develop that to 'great food choice'. This insight is therefore more complete and usable. 

Another example where you would have to be careful around the use of unigram word clouds is ‘Staff’ as the highest count.  This would mean that staff is the most talked about, but it isn’t until you dig deeper into the phrase that you may find ‘awful staff’ (bigram), ‘not good staff’ (trigram) or even the more complex phrase ‘the staff are awful’ or ‘you don’t have great staff’.  A sentiment analysis tool would be able to pull these out where a simple word cloud analysis would not.


Sentiment Analysis in summary

In summary, sentiment analysis looks at the qualitative aspects of insight. You can use tools to analyse massive amounts of verbatim feedback. Positive and negative buckets and reporting tools ensure you can get usable data from all the comments. It's not rocket science but you do need to plan your survey well to avoid ‘garbage in, garbage out’.


Whatever you do with your open ended text analysis, ensure that you really understand what it is saying before taking action. But more than anything make sure you are doing something with that valuable mine of information!


Further reading:

Read more about reporting feedback data in our guide.

Read more about extracting meaning from customer feedback data

Visit to read about our Capture Analyse Improve approach to customer experience feedback.



About author

Written by Simon Rowland

Simon is the CEO of ViewPoint. Find out more about Simon Rowland.

Leave a Reply