Customer sentiment analysis: A shift to customer service
Collaboration leverages social media
Clarabridge provides a mechanism for customer service through its Clarabridge Engage module, which directs individual messages to agents from customers using Twitter, Facebook and other social media. "The Clarabridge Suite includes many connectors to data sources that flow information into the system and sophisticated algorithms by which to analyze it," says Sid Banerjee, Clarabridge CEO. Clarabridge's technology allows scoring degrees of positivity, negativity and determination through linguistic methods of whether a comment is a complaint, suggestion or comment.
Several years ago, the goal in sentiment analysis was primarily to help companies turn qualitative data into dashboards and scores. "People were looking at reports but not at raw data," Banerjee says. "Now, individual messages that need a quick answer can be routed to the right person." The messages that are sent represent a subset of all the ones that are monitored, based on results of text analysis. That capability allows companies to build relationships with customers proactively and through multiple channels.
Another challenge in putting results into action is that even if customer sentiment feedback from multiple channels is integrated and delivered to the right individuals, different departments cannot readily collaborate to discuss the analyses. Clarabridge Collaborate, introduced in 2012, allows colleagues to discuss issues that span several departments. "In addition to delivering individual messages to agents," Banerjee explains, "we can deliver the output into a collaborative environment."
Clarabridge Collaborate can deliver alerts based on spikes (a significant change in volume, sentiment or other measured factor) or contextual alerts (specific topics or words). After an alert is sent, stakeholders can join a conversation and share their insights in an environment that documents their statements and actions in an interface that displays the data that prompted the alert.
Over time, organizations will change their structure to consolidate marketing and operations functions into customer experience competency centers, Banerjee believes.
"CRM and call centers want to use our platform to figure out what's going on," he says. "Their job is customer experience. They need to know that they can support individual customers as well as looking at long-term trends."
Although the reporting and dashboard capabilities will remain important in monitoring customer sentiment, there has been a definite pivot toward customer service within the last year. Social media is an extremely public ecosystem in which much damage can be done in a short time. Detecting a problem early through sentiment analysis allows it to be addressed and resolved quickly, to the ultimate benefit of both the company and the customer.
Redmore believes that one of the most exciting aspects of text analytics is what he calls the "convergence of silos," through use of business intelligence (BI) tools to bring together unstructured information such as social media streams to quantitative information contained in traditional BI systems. "Many companies already have powerful BI tools in house," Redmore says, "and this is a productive application for them."
Owens agrees, "The next step in using sentiment analysis and other unstructured information is to incorporate it into a broader intelligence program that includes quantitative data, and feed back this output into R&D or the supply chain."
Market booming for text analytics
When the final figures are available, the world market for text analytics in 2012 is expected to exceed $1 billion, according to Seth Grimes, founder of Alta Plana. That amount includes software licenses, service subscriptions, vendor-provided technical support and professional services. The text analytics market constitutes about one-tenth of the overall market for analytics, which includes BI software. Grimes anticipates that the text analytics market will grow at about 25 percent per year over the next few years.
Accounting for the customer sentiment portion of the market is difficult, because text analytics can be used for everything from e-discovery to monitoring social media for technical support problems and because, according to Grimes, "sentiment analysis is typically delivered as only one component of a larger social or online media analysis, customer experience or market research application." Grimes adds, "Crowd-sourcing is an alternative approach that can help businesses scale human analyses."
Many companies use text analytics platforms for multiple purposes. Grimes breaks out the search portion of text analytics as about one-third of the overall text analytics market, leaving about $800 million for the application and technology vendors. He plans to conduct systematic vendor and demand-side market analyses in spring 2013, to quantify the development of the text analytics market and sentiment analysis applications.
Grimes expects the trend to continue toward improved integration of unstructured and structured data, both from large, well established firms and more innovative startups. He also sees an increasing demand for real-time social media analytics as part of a customer engagement strategy.