It’s undeniable that word clouds were an important tool in visualizing text analysis. When compared with sifting through and reading individual comments and feedback, word clouds were definitely an improvement.
However, with the evolution of text analysis tools, the game has completely changed.” Simply knowing the most used words in customer evaluation text is no longer as useful. The words, without context, identification, or the relationship between them, doesn’t have as much meaning. So how do you go beyond the simple word clouds to get more useful information from your customer feedback, comments, and other customer-related texts? Here’s a quick overview!
1. Low Information
The number of times words appear in a select text doesn’t give you enough details to make an informed decision. A word by itself can provide you with some information, but without knowing how it was used, where it was used, or when it was used, you may end up making the wrong conclusions from it.
2. Different Words, Same Meaning
The use of different words with the same meaning in one text often results in misinterpretations when used on Word Clouds. For example, if ‘large,’ ‘huge,’ ‘giant,’ ‘enormous,’ and ‘big’ are all used once in a given passage, they would all be counted and placed in the cloud separately.
3. Context is Everything
Word clouds typically look at individual words which could be very misleading. For example, if a customer voices a bad review of your product by writing ‘Not good,’ the individual words are taken separately as ‘not’ and ‘good.’ Word clouds may then show a heavy appearance of ‘good,’ which gives you an entirely opposite perspective of your customer feedback.
4. Word Ranking Reduces Focus on Other Important Information
Word clouds work by ranking the words in a text, based on the number of appearances. The more a word is used, the ‘bigger’ it is presumed to be. For example, if your feedback data mostly contains the word ‘people,’ it will be considered as big and therefore highly ranked.
Connections and relationships between the words are not established, so there is a high likelihood of having the wrong results.
Focusing on the highly ranked words alone reduces your attention from other words that might be important.
Text Network Analysis and Visualization
Text network analysis involves representing each word as a node and every co-occurrence of words as a relation between the words. The proximity of a word to another is given a weight of connection, then the weights are aggregated in the final analysis, in a way that not only words that are next to each other have a connection. This means that even though words are not next to each other, if they appear close to one another multiple times, a stronger relationship between them is established.
Visualization of Text Network Analysis
The nodes in text network analysis represent the words, while the edges represent the relationship between them. This makes information more useful in decision making since the relationship between words and how often they have been used together gives them context. Visualization of this analysis is simply the plotting of the nodes and edges on a graph to put the analysis in a visual form that is easy to examine.
A force-atlas layout is used for the graph to push the most connected nodes away from each other and to group the smaller nodes that are connected around the center. The nodes’ size can then be arranged to the words that are most frequent in the whole text and can be narrowed down to show the words that appear most often between different topics.
Why Should You Focus on Text Network Analysis?
1. It Gives a Better Context
Text Network Analysis shows you the words used in every keyword, and how frequently they have been used, giving you a clearer picture of what your customers are trying to say to you. By grouping related words, instead of individual words, you are able to get more context of the keywords helping you make better decisions on your products.
For example, Word Clouds may show that the most used word in your customer comments is ‘expensive.’ Based on this information, your decision may be to lower the cost of your product across all packages on offer. If you, however, use Text Network Analysis on the same text and notice that the word ‘expensive’ is more connected with ‘monthly’ and ‘package,’ your decision may change, as this shows that the other packages are fairly priced, and you only have to lower the monthly package.
2. Correctly Shows the Customer’s Sentiments
Word Clouds makes it difficult to tell your consumers’ sentiments since it mostly focuses on words frequently used. Words like ‘very,’ ‘absolutely’ ‘not’ and other adjectives can give you a clear gauge of how exactly your consumers feel about your product. Failure to connect a negative sentiment to a word, can sometimes lead to you having different perceptions about your product from those that your customers have.
If one of the keywords used in the customer comments is ‘effective,’ this alone is not enough to tell the exact customer sentiment. However, if the most connected word to it is ‘extremely’ and ‘very,’ you can immediately tell that your customers liked your product very much. On the other hand, the most related word might have been ‘not.’, using a word cloud would have given you a totally different opinion to that of your customers.
Get the Most Out Of Your Customers Feedback
Your customer’s sentiments and opinions about your products or services provide you with the data needed to improve user experience. Text analysis makes it easy to understand exactly what aspects of your product they like or hate, making it easy to make effective decisions.
