Desire to Reveal Hidden Insights Driving Growth in Text Analytics

Ann All

Ann All spoke with Seth Grimes, owner of and principal consultant for Alta Plana, a company that provides business analytics strategy consulting and implementation services with a focus on advanced analytics (BI, text mining, data visualization, analytical databases, complex event processing) and on management, analysis and dissemination of governmental statistics. He is author of the report, "Text Analytics 2009: User Perspectives on Solutions and Providers," which can be downloaded free of charge.

 

All: First of all, how do you define text analytics?

Grimes: Text analytics applies statistical and linguistic methods to extract useful information from text. Text analytics can find names and match patterns and also bring out complex and abstract features: topics, facts and opinions. This may sound abstruse, but really there are very good software tools out there that make the process very doable.

 

All: What kind of growth are you foreseeing in the near term?

Grimes: I estimated 40 percent growth in 2008 to a $350 million global market, due to strong uptake in areas such as customer experience management, e-discovery and media and publishing, as well as sustained reliance on solutions in the life sciences and intelligence arena. I'm projecting 2009 growth at about half that rate, which is excellent given current economic conditions. The growth will be driven by exploding interest in social media analysis for business functions such as brand and reputation management.

 

All: Are there any other growth drivers worth mentioning?


Grimes: There's immense business value in textual sources -- conventional and social media, survey responses, contact center notes, e-mail and messaging, etc. Businesses understand they need to automate analyses if they want to keep up with this information flood. Text analytics is the way to do it.

 

All: Which text analytics applications are garnering the most interest?

Grimes: Social media analysis is red hot, for brand and reputation management and also for customer satisfaction, for which analysis of other forms of enterprise feedback also comes into play. These uses cut across industries. Sentiment analysis is key here: the ability to discern attitudes and opinions and not just facts. Quite a few vendors have jumped on the sentiment bandwagon, but not all of them are delivering great results yet. We're also seeing very strong uptake for risk, fraud and investigation applications for insurance, financial services, warranty claims and the like.

 

Some of the most interesting emerging applications are semantic search -- text analytics both enables natural-language queries and transforms raw source materials into a more findable form -- and in question-answering systems such as Wolfram Alpha. These applications will have huge impact in the coming years, particularly as we evolve toward a semantic Web.

 

All: Are there differences in what new users are interested in vs. more experienced users?

Grimes: I ran a survey last spring of current and prospective text analytics users and looked at information source by new versus experienced users.Current users are focusing on online and other feedback-rich sources. Blogs and other social media top the list at 62 percent, followed by news articles at 55 percent and online forums at 41 percent.

 

Respondents who are only starting with text analytics put e-mail and correspondence, surveys and contact center materials at the top of their lists. My conclusion: Prospective users focus initially on materials they have on hand that involve interactions with known stakeholders. Web sources can come later.

 

All: What are the biggest challenges encountered by both users and vendors?

Grimes: Getting started isn't as great a challenge as you might suppose given the availability of hosted "as-a-service" options and of solutions that are tailored to business domains such as travel and hospitality and financial services. There are also some great free, open source options that are suitable for many users.

 

The biggest challenge for vendors is to ensure ease-of-use and usability of results. User needs range from alerts to the ability to integrate analysis of data from text and database sources. You can get a lot of "lift" -- more accurate, more actionable results -- if you combine text-derived information with transactional and operational information in traditional database systems. What if you could jointly analyze online opinions and the opinion-holder's clickstream and that person's past purchases or other interactions with your company? You'd learn a lot that you could apply to deliver better products and services and boost satisfaction, sales and customer retention.

 

All: You mention in your report that most companies using text analytics are basing their ROI on increased sales to existing customers. What are other common ways of attaining ROI?

Grimes: My spring 2009 study found that top ROI measures relate to customers: sales to existing customers, higher satisfaction ratings, improved new customer acquisitions, higher retention. Clearly text analytics is seen as key to identifying opportunity and controlling risk. Organizations are also looking to text analytics for operational efficiency: to boost productivity, accuracy and responsiveness and, implicitly, to reduce costs. I was surprised that only 28 percent of survey responses linked text analytics to SEO, search-engine optimization. That figure will surely grow.



Add Comment      Leave a comment on this blog post

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.


 
Resource centers

Business Intelligence

Business performance information for strategic and operational decision-making

SOA

SOA uses interoperable services grouped around business processes to ease data integration

Data Warehousing

Data warehousing helps companies make sense of their operational data