It is commonly estimated that 80% of enterprise data exists as text and other unstructured data. Text is everywhere – in emails, documents, reports, forms, call logs, press releases, blogs, tweets and more. Text analytics, or text data mining, is the process of automatically extracting concepts, topics, facts, sentiments, relationships and other information from text. It can be used to classify documents, automatically route emails to the right person, identify specific quality or service problems, monitor the competitive marketplace and better understand the “voice of the customer.”
While text analytics is clearly much more difficult than quantitative analysis, it can provide insight that is even more specific and actionable than traditional quantitative BI data. It also offers the potential to be able to respond in real-time to customer comments because of the speed at which computers can process text information. But the greatest insight will come from combining both text and quantitative data mining results together, often in combination with predictive analytics, to identify issues, take specific actions and improve business results. Here are a few ways that companies have gained insight through text analytics.
- A large global hotel chain analyzed customer service call logs, letters, comment cards, surveys and emails. A manual analysis that took days identified lobby cleanliness as a problem at a specific property. Automated text analysis on the same data took hours to identify that a dirty carpet in the hotel lobby was the source of complaints.
- Another hotel chain discovered that, in large resorts, guests who were looking for a location at the resort had a much better customer experience when the bell staff walked them to the location rather than just pointing them in the right direction. These are both good examples of the kind of specific, actionable information that can be extracted from text data.
- A consumer product company used text analytics to understand what customers were saying about their product in surveys and social media. They mapped Twitter handles to their customer database to better understand why some customers are brand promoters and others are brand detractors. They were able to discover new information about what is most important to customers and why customers use their product. The information gleaned from text analysis led to changes in product development, marketing and customer service.
- A telecommunications company knew from quantitative analysis that many customers called prior to canceling their service. By mining call logs and combining that with customer information, the company was able to identify the specific questions asked by customers who were likely to cancel their service and to take action to try and prevent cancellations.
- ConAgra has hundreds of work-from-home call center agents who record customer comments verbatim. The company mines these comments in real-time to identify and act on trends. When a number of comments appeared that a particular product had too much vinegar, the company was able to track down the specific plant that produced it and take corrective action immediately. When some customers began complaining that their ReddiWhip was watery, ConAgra identified that a certain grocery chain was not refrigerating the whipped cream properly and they were able to resolve the problem.
Subscribe to our blog to stay informed on text analytics and other data analytics topics.
Spotfire Blogging Team