June 1, 2011 – Most banks are still formulating their overall social media strategies. But many are nonetheless monitoring and archiving customer postings they say will inevitably inform the development of predictive models, which they'll apply to social media to increase revenues, as well as improve customer service.

Bank of the West, for instance, has started to build social media models to increase new account openings, expand product additions and enhance services. The next step for the San Francisco-based bank is to optimize specific, targeted offerings aimed at high sales conversions by more precisely discerning the value of social networked connections between the bank and its customers. The firm wants to build models that can take advantage of any sweet spots where opportunities are easily converted to revenues.

"We'll do lots of different, smaller-scale models first to try to understand what the implications of social media are," says Joel Kleinman, vp of internet channel business intelligence and analysis at Bank of the West. "We're trying to understand the predictive value of knowing that a visitor came from a social network. But what we're ultimately trying to do is translate the value of the bank-customer relationship into monetary terms."

The bank is taking statistical lessons learned from previous campaigns involving social networks to develop new offers. To generate revenue and establish new relationships during the rollout of its mobile banking application last year, for instance, Bank of the West gave a free iPod touch to anyone who signed up for a new checking account and tapped either direct deposit or bill pay services.

After the first several weeks of the campaign, the bank found that social media sites like Facebook and blogs were the largest referral sources for customer acquisitions. On the last day of the campaign, such sites corresponded with what the bank believes was an outsized number of new online account applications filed, due to the networked relationships of participating customers.

"So what we saw was a small blip in Facebook equaling a much larger blip in acquisitions," Kleinman said. "We couldn't directly tie every instance to a Facebook posting, but the relationship was certainly there. The lesson for us is that social media has an influence on how it affects the bottom line, if there's something we're doing that lends itself to networked marketing."

The bank has since found that about 38 percent of the variation in its online account openings can be attributed to direct referrals from social media sites. The firm marks a similarly significant referral rate to its branch locator page from such sites.

"That would indicate that social media is driving highly qualified traffic to our pages in terms of how interested prospects are in establishing a relationship with the bank," Kleinman says. "The links we've seen between account openings and direct referrals to our Web site from Facebook or Twitter are at least strong enough to warrant more efforts."

Bank of the West is using Web-facing vended systems to track and categorize social media and other data, including Marketwire's Sysomos to monitor social networks; Adobe Omniture's SiteCatalyst, Test and Target, and Discover, to analyze the bank's Web site activity; and ForeSee Results to track customer survey feedback, where there's a lot of unstructured data.

But the predictive analytics the bank is applying to this information has mostly been developed in-house and with staff assistance, although the bank uses IBM's SPSS to optimize cross-selling and Teradata for enterprise analytics.

The majority of banks, like Bank of the West, remain in "a listen and learn" mode; talking to vendors, collecting data, tweaking models, trying to figure out just how they can include themselves most effectively in social networks.

The way Royal Bank of Canada envisions deploying predictive analytics to social media "in practical application, would be to have a conversation with a customer or prospect on Facebook," says Michael Wong, director of enterprise business intelligence at RBC. "That's part of our marketing initiatives: If there's a conversation that we believe we can perhaps engage in and provide benefit or prospects, we'll step in to provide advice."

RBC uses its own, in-house built predictive analytic tools "utilizing various statistical modeling and advanced analytics technologies" to tap areas ripe for "revenue growth, and to try to identify risk and opportunities," Wong says. RBC uses SAS as an enterprise business intelligence tool, but it's fed with the bank's internally developed models built within specific business areas like marketing, contact centers and fraud prevention. (For more on RBC's client data efforts, click here for the "25 Top Information Manager" profile of Mohammad Rifaie, VP of enterprise information management.)

Most banks have for years gleaned intelligence from such semi-structured text, or data that's not sub-structured in a table format from incoming customer contact channels. But expanding such modeling to social media has posed multiple challenges.

For instance, it's still not easy to link an online persona from Facebook to an actual client. RBC's Wong points out that "internally at the bank there are eight different Michael Wongs; on Facebook there are probably hundreds. We need to understand any privacy concerns when tracking online activity."

Lucien Randazzese, senior director in TowerGroup's specialized advisory group, says "banks are not anxious for their customers to know they're trying to use this data to sell them, say, a car loan."

Institutions also risk violating public advertising and communications rules. "Banks are deathly afraid of million-dollar fines they could possibly incur just for being flip on social media sites," says Neil James, digital analyst at Minneapolis-based PR firm Russell Herder.

Some of the reluctance stems from structural inadequacies. "Banks predictive models are set up to make offers when a customer presents him or herself to call center or branch personnel," says Ron Shevlin, senior analyst at the Aite Group. "How is that going to work in social media? Conceivably, a bank can identify a customer on social media through cookies. But they'd still have to integrate the results of the predictive models created by an enterprise CRM application with the social media channel: That's easier said than done."

Lack of historical data can also pose problems. "A lot of the sites tend to remove a lot of content," Wong says. And James says about 40 percent of incoming mentions of a company's name that are picked up by top-shelf social media monitoring programs are the result of spam or affiliate marketing.

This story originally appeared on Bank Technology News.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access