Text analytics has wide application in a business environment. For users wishing to reap productivity gains, text analytics can automatically filter spam, categorize messages (emails for example), label documents and enable fast and relevant information search. This can be viewed as the ‘traditional’ role for text analytics, although more contemporary applications include fraud detection, warranty claim processing, competitor analysis and social media monitoring. Beyond this text data are being used to add new features to data mining activities with the aim of creating predictive models which might be used to detect possible customer defection, new selling opportunities and delinquency. It can also be used to provide new insights into customer behaviour through identifying archetypes.
The potential benefits of text analytics are not ‘automatic’. Domain experts are needed to provide input to the analytics process and sanitize results. This applies to almost every application of the technology regardless of whether a ‘plug and play’ type solution is being used or a bespoke application has been built. Even web based services which provide sentiment analysis of social media require considerable amounts of configuration if results are to be meaningful.
The focus for text analytics solutions is primarily in the customer and marketing domains. Such solutions are often cloud based, but for larger organizations a in-house deployment might be necessary because of latency and security issues. Either way text analytics provides insights into customer behaviour that are not accessible through analysis of the structured data contained in databases. This extra dimension can be used to tailor a more relevant interaction with customers and predict future behaviour. For example it may be possible to identify which customers are perfectly good credit risks, but sometimes make later payments because of lifestyle. It is much more likely that a customer ‘sweet spot’ can be identified through text analysis than by any other mechanism, since text contains significantly more potential information than the history, demographics and other data held in databases.
In a marketing environment, text data in the form of open surveys (where respondents can add free text comment), can be used to extract nuances which simply cannot be accommodated in a closed response form. This might enable sentiment scores to be created or the identification of terms and concepts that had not been anticipated. Obviously this is closely related to the sentiment analysis of social media, which at the current time is over-hyped, but is quickly evolving to provide behavioural insights and trend analysis for marketing activities.
While customer and marketing applications might be the most obvious ones, text analytics applies to any domain where text data is acquired. In a manufacturing environment the notes made by maintenance staff might be used to improve the prediction of maintenance requirements and the avoidance of down time. In a medical environment text notes that are captured during the diagnostic process can provide valuable input to understanding patient concerns and the process of diagnosis itself.
Perhaps the most promising application of text analytics is the creation of new features for data mining processes. Combining structured and unstructured data in this way facilitates the meeting of two quite different information ‘dimensions’ and is already being used in sales, marketing and customer care applications.
As always business management will be tasked with the need to identify opportunities and decide how unique they want a solution to be. Packaged solutions potentially reduce risk, but also reduce opportunity. A bespoke solution introduces technical and business risk, but also provides the most opportunity. Fortunately a number of suppliers offer a middle way with many of the technical, architectural and business risks largely mitigated, but with an opportunity to deliver a tight fit with individual business needs.
Next article in this series: Text Analytics Strategy
Text Analytics: a business guide
A report for business and technology managers wishing to understand the impact of rapidly evolving text analytics capabilities, and their application in business.
The Business Value of Text Analytics
What is Text Analytics?
Text Analytics Methods
Unstructured Meets Structured Data