Community management specialist uses known users to validate its analysis of others.

When Lithium Technologies talks about social media monitoring, the conversation always comes back to community.

Social media monitoring and analytics is a hot sector, but also an increasingly crowded one. Tools to spot social media mentions of a brand or company and then try to divine the positive or negative sentiment behind those comments have been in the spotlight lately, with Salesforce.com announcing the acquisition of Radian6 in March. Taken together with Jive Software's acquisition of Proximal Labs in April and many other deals and product announcements, social media is looking like the next analytic frontier.

On the other hand, based only on the claims in their press releases, you might think these services were all alike--all claiming to be able to digest the entire Twitter "fire hose" of social chatter, along with blogs, blog comments, and public Facebook content. They then run all that through keyword filters and natural language processing software and spit out a list of posts and tweets that you ought to pay attention to.

"We can do a pretty typical 'share of voice' and sentiment analysis on those data," Lithium CTO Michael Wu said in an interview. But if elements of the social listening analysis are "pretty typical," what Lithium allows you to do with the data is not, he said.

"The thing we offer that's a little bit different is we think listening is really the first step. You want to listen first to gain some insight, and then take some action. Otherwise, it's like telling you tomorrow the world going to end, but you can't do anything about it," Wu said. In Lithium's view of the world, one of the most effective ways your brand can take action is by building a community of customers who can help you respond to a volume of questions or criticisms that might otherwise overwhelm whatever service or support team you have assigned to the social channel. Lithium also uses its access to a relatively manageable community of known users to fine-tune its understanding of the broader communities on the web, Wu said.

This focus on community reflects Lithium's core focus on software and services to help its customers operate their own online customer communities, with a particular focus on providing social incentives and "game dynamics" like rankings of the most active members. By rewarding active participants--often with nothing more than an ego boost--Lithium's corporate customers aim to replicate the feat of Tom Sawyer convincing the neighborhood kids to paint Aunt Polly's fence. With community members helping each other or pointing newcomers to community resources like frequently asked question documents, these companies try to magnify the effectiveness of their customer service teams.

Lithium CEO Lyle Fong said one of the keys to doing this effectively is identifying the community members who truly are product experts and evangelists. "We can tell you, 'Here are the 20 users who really matter,' " he said. Because Lithium hosts these web communities (which can be set up to appear as a subdomain of its customers' websites), "we have watched over 10 years to see what has worked and what has not worked," he said.

These privately managed communities bring some of the social conversation into a more controlled environment, but that doesn't change the fact that lots of other conversations about the brand may be going on elsewhere on the web. One way of integrating these communities with the broader universe of social media websites, blogs, and message boards, it to give community members the opportunity to see some of the questions and criticisms gathered with social media monitoring and encourage them to respond.

Wu believes that may be the only way a consumer products company can address the scale of social media. "The company can never hire enough people to respond to everything happening in social media," he said.

Lithium faces the same challenges as other social media analytics companies in trying to make sense of social media comments, often expressed in a language of abbreviations and slang, where automated systems often miss the subtleties of human language. Wu's example: a movie review that says "read the book" is negative, whereas a book review that says "read the book" would be very positive, but a sentiment analysis system would tend to score them both positive.

On the other hand, when it comes to other measures like determining how influential a blogger or Twitter user is, Lithium has an inside track because it operates its own communities where it has more detailed information about participants and their interactions with each other, Wu said.

Even though Lithium can't rely on the same level of detail when profiling Internet users outside its community, it has constructed its model of who is the most influential using metrics of followers and connections, as well as their level or participation. It then checks and refines that model by applying the same analysis to own communities, initially without reference to its inside information, and then reconciling it with the internal metrics.

"The community provides us with an independent source of information that can be validated," Wu said. And that should ultimately lead to a better analysis of the social web as a whole, he said.

Want to improve effectiveness and reduce costs by becoming a more social business? This webcast from InformationWeek and BrainYard can help. It happens May 18. Find out more. (Free registration required.)