vocCustomers frequently post their complaints on web portals, forums and  social media in form of free text. Analyzing this data is very important in a sense that it gives us useful information into issues facing end consumer on basis of which company can adjust or improve it’s business process. Any such tweaks should result in reduction of complaints regarding particular process.

Let’s look at small fraction of consumer complaints for a top 3 Travel aggregator in India.

  1. goCash not received
  2. CANCELLATION REFUND NOT RECEIVED  
  3. Promo Code is not working 
  4. Reschedule 
  5. poor customer service 
  6. Contact no for Hotels 18012081059 is invalid 

These are just subjects of complaints with further details of the problem not quoted here. Each of these complaints serve as important pointer to consumer problems. One simple solution that provides bird’s eye view of what customers are speaking about is word cloud. We took last 2 months of this data and ran it through text mining algorithms.

ccvoic

consumers have been frequently complaining about in last one month and two months. Insights can be further improved by data cleaning, building dictionary of industry specific stop words and stemming. This however is just the tip of what we can do.
Does your support team needs to read long paragraphs to decipher what customer is talking about? How about if we could identify themes from these paragraphs that can point us to what exactly consumer is talking about? We won’t have to read all those lines. You could prioritize addressing consumer complaints. This is the realm of Text Mining, Natural Language Processing. We will explore more about themes extraction in our next post.
Tags:

Leave a Reply