The exploding social media environment is driving businesses crazy. Figuring out how to mine the unstructured information about their companies and brands on the ‘Net-and what to do with it from there—is among the top two or three projects in companies, we hear.

This brave new world of ever-growing customer interaction today has led to the acquisition of Biz360, a social monitoring and market intelligence vendor, by Attensity Group, whose technology helps companies process and analyze free-form text using natural language processing (NLP), machine learning, artificial intelligence, and semantics. Given its role in helping companies understand their customers, sentiment analysis is of course a hallmark of Attensity’s technology, through its Exhaustive Extraction technique. The new world we’re in also has recently helped fuel the launch of SAS’ Social Media Analytics hosted service for helping companies understand, predict and act on social media data, including sentiment. Other news in the emerging Social Customer Relationship Management (Social CRM)/ Voice of the Customer/ Customer Experience management space includes Clarabridge’s enhancing its Enterprise 4.1 text analytics software with features including intensity-based sentiment analysis. For its part, Expert System recently published its findings on how Toyota, using its Cogito Monitor solution, may have been able to be ahead of the damage control game if it had been monitoring customer sentiment online – and there was a lot of it. From Feb 2009 to Aug 2009, it reports, Toyota was one of the most talked about brands, with almost 4,000 posts during this period — and that is just among the most popular U.S. auto blogs and forums.

It’s that massive scale of social data out there across so many different sources – Facebook, Twitter, topical forums, etc. – that can be terrifying to marketing and other departments charged with courting, managing and servicing customers. For big companies with lots of consumer customers there could be tens of thousands of posts made daily about their products or their brand, and that drives questions about how to pinpoint not only what department but who within that department should know about particular comments, how can that be communicated efficiently, and responded to effectively. “Social media has created new rules of the road for engagement,” says Michelle de Haaff, Attensity CMO. “Providing a ‘thank you for your comment’ is not okay. You are looked at as a company that has institutionalized social media when the point of social media is for humans to interact with each other.”

The acquisition of Biz360 is key to this, she says, taking the Attensity portfolio beyond its roots in supporting internal customer data understanding and analysis (surveys, emails, repair notes, etc.) and particular site analysis to a deep and broad listening capability across the entire social web, and from there bringing the value of online conversations to specific business processes. Biz360 covers over 50 million online sources from traditional blogs to opinion media (like Amazon reviews) to social media. The vision, which Attensity is calling Engagement 2.0, is to provide a complete platform for listening and analyzing these conversations, automatically directing them to the right sources within a company (thanks to NLP understanding what the conversation is about), suggesting to those sources the appropriate response, and delivering the right information to the customers asking for it online. So, for instance, if someone online says she thinks the iPad is cool and is considering buying it, such a message would go to someone in Apple sales. From there the technology’s role would be to alert a salesperson about the appropriate links to deliver to the customer for more information, which he could transmit with a personal message about how he hopes she’ll ultimately purchase the product.

(Photo: Christophe Goessen/Flickr)

Delivering more complete automation of this cycle is a little more future than present, but de Haaf says it’s Attensity’s hope to apply natural language and semantic tech and build around Biz360′s features to this end, in a product that will be renamed Attensity 360. Similarly, social knowledge can be communicated back to the corporate knowledge repositories Attensity supports to support other customer communications cycles. Social media data, emails, call center notes and other conversations can all be accessible for a 360-degree view of what customers are saying.

Data Here, Data There, Data Everywhere
That’s important, because as massive as social media is, it’s just one link in the overall chain of customer data.

Such integration among all these different sources of customer input as Attensity mentions is important to the bigger picture of acquiring and understanding information and sentiment coming from consumers, and acting on it. It is also at the heart of what some other vendors are doing. With its Social Media Analytics service, for example, SAS is analyzing the unstructured data on the social web, as well as structured information — building on its text analytics heritage — and integrating it with CRM, marketing and other systems that may host call center logs, surveys and other customer-infused data. The foundation of the solution is the capture and integration of all the data into an underlying warehouse, bringing text and sentiment analytics to bear on top of that, and then providing interfaces for customer interaction. This includes a web portal with dashboard information, and a behind-the-scenes interface on how specific topics are being scored for sentiment at a document level, an individual blog post level, and even some of the words that could be driving those scores.

And it allows for adjusting rules for modeling data and reprocessing everything to use those new set of rules – e.g. changing ‘wicked,’ which generally has a negative connotation, to reflect that it’s a good thing when it comes from a poster in New England, where we’re told it predominantly is used as a positive word.)

The on-demand platform provides clients who sign on to the service two years of data from day one to help in trend-spotting. “Step one is really understanding and delivering insights that in and of themselves are actionable,” says John Bastone, global product marketing manager, customer intelligence. “If the platform just showed positive or negative sentiment for a brand, that’s useful but what do I do with that?” What makes that information live are being able to do things such as identify, for instance, current vs. past perceptions, and how key influencers might be singing a different tune about, say, hotel cleanliness than they were for a similar length of time three months ago. And take that to the next level to compare what is being said externally on the web to what is being routed through various internal CRM systems.

“The internally sourced data is another source of conversational data that can be modeled and mapped to the same concepts in the social paradigm,” he says. At the end of the day, everything is rooted in the approach of integrating data, mapping it to what companies care most about in terms of issues, building out those concepts by ripping through hundreds or thousands or even millions of documents and blog sites, reviewing them, and then applying sentiment and letting people explore that in every possible permutation imaginable. Then that information has to be accessible to business users easily, which is where the web-based dashboard, reports and workflow-enabled alerts come into play in the SAS on-demand solution. “Because we can model data as granular data as it relates to specific business issues, this lends itself well to action,” Bastone says. “So the ability to set up alerts, to drive certain workflow actions in an enterprise context is something we fully expect to bring to some of our larger clients.”

When it comes to text and sentiment analytics around customers’ input, Clarabridge CEO Sid Banerjee makes the point that you need to be comparing and contrasting these insights, because you might learn very different things from them. “You often find that marketing concepts spike in some areas and service and quality issues spike in other data sources,” he says. “They don’t always spike at the same time.”

Another thing to keep in mind, Banerjee adds, is that technology such as his company offers will do the work of routing spiking issues to an individual within a company that has accountability for it through alert-based mechanisms — but it would generally be difficult to dictate business rules to automate all responses. Suggestions can be made about next steps that an individual can act on, but there can be a lot of different ways to do something with the insight you glean from monitoring social media and other customer input for sentiment, attitudes, and the like.

“In the world of quantitative decision-making you can define a decision rule — if you are a store running out of something and selling this much a week and have x weeks of supply on hand, that’s fairly straightforward and well-proven method for ordering. But qualitative analytics, where we are, is different. How do you use qualitative insight? People haven’t decided yet that certain qualitative insights dictate quantitative outcomes yet. Are we moving there? Maybe.”

The Cloud is Needed
One thing you’ll notice about all these offerings is that they are hosted platforms. There are a couple of reasons. One thing, Banerjee says, is that a lot of the primary buyers of platforms that deliver text and sentiment analytics around the customer, especially the social media user, are business-people in customer optimization, call center, or product marketing and management channels. They’re not IT people interested in heavy technical integration or with a lot of analytics background. “They just want something to work, to be accurate and scalable,” he says. Just 25 percent of its business these days is on-premise.

Bastone has seen the same thing. “A lot of the groups that were coming to us were PR or Marcom or Sales or Warranty. They weren’t necessarily the groups that had a tremendous amount of analytics experience and almost always weren’t groups that had any kind of text analytics or text mining expertise,” he says. They understand the concepts but don’t have the staff in place that had skill sets to leverage that solution, to figure out how to implement data integration, apply text analytics and mining on top, and then hook in sentiment analysis and then build a BI stack on top of that for reporting.

The other reason is the sheer volume of monitoring underway. “The ability to listen to massive amounts of conversations that take place online requires massive, massive scale,” says de Haaff. “To do it in a scalable way requires a cloud/SaaS environment.”

Announcing Semantic Tech & Business Conference - San Francisco 2012

Semantic Tech & Business Conference is returning to San Francisco in June! Join us from June 3-7 for complete coverage of Big Data, Linked Data, Extreme Information Management, and Semantic Web. From breakthrough approaches to solving business problems to the big data implications of fast–evolving technologies, SemTechBiz provides you with an unparalleled interactive experience and delivers tangible business value. We're offering a special early rate when you register by February 17. Sign up now!