How Sentiment Analysis Can Help You Hear the Voice of the Customer

For companies, it is critical to understand the feedback that customers are giving them through all sources.  Nowadays, lots of customers are expressing their opinion about products and services in unstructured data on social media and other sources, such as Facebook, Twitter, product reviews, blogs, free text notes on customer call center interactions, surveys, etc.

The approach that best captures this feedback is Voice of Customer (VoC).  Traditionally, VoC is the collection and analysis of customer wants/needs, preferences, and expectations.  It covers all interactions that a customer can have with a business.  The introduction of massive amounts of unstructured data via social media and other new means of communications offers new challenges to VoC – how can it effectively capture the critical information in this data?

Capturing such data is hard to do because too much noise is drowning out the valuable information in unstructured data.  We need tools to improve the analysis over very large amounts of data because manual sifting is not just time consuming and resource intensive, but is also basically infeasible.  There’s just too much stuff.

NetOwl’s Sentiment Analysis automates the VoC process and provides insights that can’t be gained by manual means.  It analyzes how customers are perceiving each of the different kinds of interactions with a company:

• NetOwl can analyze complex sentiments: “The hotel had a quick check-in, but room service was awfully slow.”  NetOwl understands that there are two sentiments contained here and returns a positive sentiment for the check-in  but negative for the service.  This is critical for getting down to the fine-grained analytical level necessary for a corporation to understand where it’s making the grade and where it’s failing.
• NetOwl combines its Entity Extraction with its Sentiment Analysis to identify the specific opinions customers have about a corporation’s products and services: “I hate XYZ Corporation’s refund policy on its new smart phone.”  NetOwl understands that XYZ Corporation is a company and that the refund policy is XYZ Corporation’s.  It identifies that XYZ Corporation is the manufacturer of the smart phone.  This is the necessary underpinning to understand the sentiment being expressed in this comment.
• NetOwl normalizes all sentiments expressed in variable language into one format based on a unique ontology that will allow aggregation of sentiment contained in very large numbers of comments, opinions, reactions, etc. to power the in-depth analysis required. This normalization also allows the combination of information coming from unstructured sources with that coming from more traditional structured sources.

Understanding what the customer is saying is very difficult in a crowded and noisy information space like that of social media.  Sentiment Analysis is necessary to automate the otherwise painful, slow process of wringing true value out of unstructured data.