View the original post.
While the overarching goal of content analytics–to capture digital content and then apply business intelligence (BI) in order to glean actionable insights–has stayed relatively fixed in recent years, the continued proliferation in variety and volume of content means that both vendors and users of content analytics solutions must move at a brisk clip to even stay in place.
In a 2015 industry survey called “Content Analytics: Automating Processes and Extracting Knowledge,” conducted by AIIM (Association for Information and Image Management), the majority of the 238 respondents felt there was real business insight to be gained if they got content analytics right. Doug Miles, AIIM’s chief analyst, says, “We take our input from users rather than vendors or other analysts, so what we see are much more general trends rather than great excitement. However, while 17% of survey respondents consider content analytics to be an essential for their organization now, 59% feel it will be essential in 5 years’ time.”
So what’s happened in the past year to make this essential capability easier to access-and how will organizations move the ball down the field?
The Year in Review
Boris Evelson, principal analyst serving application development and delivery professionals at Forrester Research, says that when it comes to text analytics, the changes have been mostly incremental. “Yes, data sources are getting broader and deeper, but core components of text analytics such as data preparation, natural language processing, linguistic rules, statistical analysis, etc., haven’t really changed that much,” says Evelson, who co-authored Forrester’s November 2015 report, “Vendor Landscape: Big Data Text Analytics.”
The most important development in analytics, according to Meta Brown, a consultant, speaker, and writer who promotes the use of business analytics, is simply that new people are becoming aware that it’s possible to do something with it. “People are exploring what they can do with analytics beyond just making summaries,” she says, adding, “They’re probably less aware of what’s possible with audio and video content.”
Online search remains the place where most people see content analytics in action. Brown says, “If someone is searching and their search term doesn’t match the terms used internally, organizations are using content analytics to start bringing those things together,” so instead of no answer, they get a search result. AIIM members bear out the notion of search as a logical starting point, says Miles: “The response to our survey indicates that improving the effectiveness of search is the first priority of most users.”
Combining content analytics with other types of data-both structured and unstructured-has been a fertile area for growth. Thanks to trends in Big Data, the complexity and scale of that challenge have put new emphasis on cognitive computing, which strives to use data mining, pattern recognition, and natural language processing to solve problems in a human-thinking way. But Evelson points out that cognitive computing solutions such as IBM Watson aren’t a quick fix. He says, “They require so much preparatory work and so much investment in human resources-like data scientists, linguists, and subject matter experts to train the system-that the real ROI of cognitive computing is still TBD.”
A Look Ahead
Even as new content analytics tools come to market, there is still a significant context gap. Brown offers the example of a user performing a work-related search on a home computer, only to have work-related advertisements follow her into her leisure computer time. “A major technical challenge exists in matching relevance,” she says. “Vendors say they’re offering a 360-degree view of the customer, but knowing everything about a person doesn’t actually tell you anything about the person.”
Brown says it’s better to think about the real-life interactive counterpart to the online behavior and model analytics solutions to mimic them. Instead of a salesperson approaching a shopper to discuss a product he was browsing three stores back, she says, “I walk into a shop and they ask me questions relevant to that specific interaction and combine it with their own knowledge about the product.” Enterprises are going to continue to seek solutions to provide that sort of real-time, context-rich understanding of what problem the client is trying to solve in that moment.
Evelson also sees the context question looming large: “One trend I’ll be watching is what Forrester describes as ‘insights to action.’ Most text-analytics vendors can extract customer sentiment. But what does it really mean in terms of how a vendor should then treat that customer? Send them an offer? Offer a discount? Drop them from the prospect list? I am seeing some inklings of the next generation of sentiment analysis which also attempt to predict emotion and more importantly intent. At the end of the day, it’s not the sentiment, but [the] intent that’s important in order to focus customer treatment.”
Related to the context question is how organizations ensure that content analytics achieve positive ROI. The financial services industry has probably made the most strides in using content analytics to support real-time decision making, but it’s becoming more prevalent in other functions and industries as well. “They’re trying to make customer service use analytics more to make sure that inquiries are automatically routed to the correct place,” says Brown. Tech support is another arena getting a boost from content analytics, such as when users begin to type in a question and answers populate in real time.
Miles agrees, saying, “The ability to steer inbound content and to set the workflow through the process or to drive compliant case management seems to offer a somewhat more pragmatic return than many of the more ‘exciting’ business insight applications.”
As the content component of Big Data becomes even more diverse, look for demand to build for better integration of text analytics and structured data analytics. Evelson says, “Text analytics tools are great at extracting structure from unstructured data but fall short in comprehensive analysis of the results. BI tools are great at the latter task, but do little with raw unstructured data sources. Bridging the gap between the two will be a priority in 2016.”