Making Use of Unstructured Feedback from Customers
Many customers want to tell you what they think and how they feel. Are you really listening them? Most likely you are saying ‘YES’ and you’re not alone! It is fair to say that companies have become better at listening to their customers in recent years. Yet, how deep do you dive into your customer feedback? Are you just looking at your star ratings coming from surveys or are you looking at your customer comments incoming from anywhere, at any time and in any format?
Do we really need all these customer feedback?
According to a recent study by IBM, over 80 percent of all incoming feedback from customers is now in an unstructured format including emails, online reviews, social media posts, comments in surveys, recorded phone calls, etc. Understanding your customers satisfaction and expectations without looking at all these customer feedback is like trying to complete a puzzle with missing pieces. If your company is getting a lot of comments and reviews to read and manage, it is certainly a challenge to handle them.
Getting insight out of unstructured feedback
While structured data which comes through surveys can easily be analyzed by applying simple statistical methods, unstructured data is significantly harder to understand for a number of reasons such as allusions, multiple unrelated ideas, grammatical errors or misspellings, unclear sentiments, etc. For example, it will be a challenge to understand a customer who says “I thought the software would be useful, but it was a Trojan Horse” unless you know about the Trojan War in mythology!
The simplest way to understand unstructured customer feedback is simply to have trained staff read all the comments and categorize them. If the company is receiving a few hundred surveys a week, this approach may suffice. However, it does not scale well when numbers grow for several reasons such as limited staff resourcess, human error possibility and high cost. Alternatively, text analytics solutions provide an automated way to read through all of your unstructured customer feedback to determine which topics are on your customers’ minds and to trigger improvement actions when needed.
How action oriented text analytics works
Action oriented text analytics can sometimes be confused with sentiment analysis. They are both ways to derive meaning from unstructured customer feedback, yet have quite different purposes. Sentiment analysis classifies if an expression is positive, negative, or neutral. For example, “The hotel is in a great location with fantastic view of the city” has positive words “great” and “fantastic.” Now consider a similar sentence: “Fantastic location and great view, ya right.” The same positive words appear but the customer is using sarcasm to make a negative point. Most often, sentiment analysis using simple techniques may not catch this. On the other hand, action oriented text analytics, through the use of natural language processing (NLP) and machine learning techniques, identifies the parts of a text, learns which words and ideas are linked, automatically corrects for mistakes and transforms them into useful business intelligence upon which improvement actions can be taken.
As an example, through the use of a text analytics solution, an automotive company may learn whether the customer(s) is talking about test drive interaction that he/she recently had at a particular dealer, whether the problem was insufficient information given or the drive was too short, etc. Then the solution automatically notifies and triggers improvement task to the related dealer manager to reach out to the customer for corrective action while taking preventive measures for the future.
Technological advancements in text analytics capability has made it possible for many companies to automatically process and take action on millions of unstructured customer feedback at a much smallar cost and much faster!
For further information, please contact Müzeyyen Kaygısız at firstname.lastname@example.org