Busting down the info silos

Savvy business leaders are starting to recognize the paybacks of helping all their business groups work from the same data

More than a decade ago, General Mills developed a CRM system called International Contact Entry (ICE), which collects consumer contact data from phone calls, mail, email, and the Web and puts it all into a single Oracle database and a SAP BusinessObjects reporting system.

"We now have a single database for global consumer contact information that spans the various silos of division, plant or country," says Jeff Hagen, the food giant's director of consumer services. The business analytics group develops dashboards and reports to serve not only his group, but also the Quality and Regulatory Operations (QRO) group, whose job is to ensure product quality and safety.

[ Discover what's new in business applications with InfoWorld's Technology: Applications newsletter. | Get the latest insight on the tech news that matters from InfoWorld's Tech Watch blog. ]

This was only the start of Hagen's decade-long voice of the customer (VOC) crusade to provide different business groups with the customer intelligence they need. A few years ago, he convinced management of the value of giving sales and marketing access to customer intelligence, through ICE.

Here's one example of why breaking down the silos paid off: A product locator on General Mills' website allows customers to find out which stores in their ZIP code area carry, say, a particular Yoplait flavor or type of cereal. They can also enter which store they prefer to shop at.

Hagen's group compiles and analyzes such queries to determine which products consumers are hunting for the most, and which they are having trouble finding in any given region. Salespeople then use the reports to convince store buyers to purchase more of those products.

Another example is an early warning system that lets General Mills' QRO group know when customer complaints about a particular product begin to escalate, so they, as well as consumer services, can deal with the situation quickly.

A couple of years ago, it became clear that General Mills needed to extend its VOC platform to include social media. More and more business groups were signing up with social media data mining services, creating silos of valuable information that only they could access, Hagen says. "We wanted to provide a single pipe for all [social media] conversations for all our brands, not just for groups that could afford to do it themselves," and ultimately make that data sharable, he adds.

Meanwhile upper management, alarmed at well-publicized viral disasters that had befallen other firms, wanted Hagen's team to build a system that would "enable us to keep our finger on the pulse of social media," says James Bell, an IS staff consultant who works with Consumer Services and developed ICE.

Last summer, the company deployed the system, which, combined with ICE, "allows us to provide our brands' teams with a more complete picture of the desires, needs, complaints, praise and frustrations of our consumers," Hagen says. Still, the VOC initiative is still very much a work in progress, he admits.

Beware, though; this type of project isn't meant for those who are watching their pennies. Hagen says General Mill has invested in "the hundreds of millions" of dollars in these systems "over the years."

 Silo-less living

General Mills is hardly alone. Savvy business leaders are starting to recognize the paybacks of enabling business groups -- ranging from marketing and sales to quality control and engineering, to research and education -- to work off the same data. Another goal is to help business users to share and collaborate on the reports and ideas that data generates.

Enterprises started merging data from different business systems decades ago, starting with the different modules in their enterprise resource planning software, says Bill Gassman, an industry analyst at Gartner. "Now we're in that cycle again," he says, but with a difference: The data is being collected, processed and generated by business groups and end users, often without IT's involvement or even knowledge.

An online survey of 350 corporate decision-makers, commissioned by digital marketing vendor DataXu and released in March, found that only 15% of respondents in marketing involved the IT department in marketing technology purchasing decisions.

While end-user-generated "information stovepipes" can provide great value to business users, they constitute a major data integration challenge, says Gartner's Gassman. This is particularly true of social media data, he adds, because it exists in so many different forms, many of them unstructured.

Vendors have started to address these needs only recently, and in a limited and often proprietary fashion, business sources complain. Respondents to the DataXu survey, for example, identified the lack of a single, cross-channel digital marketing platform as the number one obstacle to the growth of digital marketing.

The technology involved

Integrating all the silos represents a two-pronged challenge for IT executives, says Holger Kisker, a principal analyst at Forrester. On the back end, they need to extract data from multiple sources, transform and, in the case of social media, structure it, before loading it into a cohesive infrastructure. This could be a data warehouse, or a virtualized structure in which data remains on distributed systems.

Often the best strategy is to "bridge rather than flatten" the information silos -- in other words, use APIs or some other source to connect the data sources rather than try and load all of the data into a single system, Gassman advises. "You'll never get rid of silos, they're too valuable," he adds.

End users are building "stovepipes" to social media sites; this allows users to access information "that IT has neither the skills nor the resources to provide," Gassman says. The value of such pipelines greatly increases as businesses engage their customers via social media, he adds.

For example, marketing departments are increasingly launching ad campaigns on social media sites, which can then be monitored for consumer reactions. A stock trading company discovered, through monitoring, that people who participate in its online social community average 360 trades per year, versus 200 trades for those who don't, Gassman reports. "So they're encouraging customers to participate in the community."

Connecting all the stovepipes into one big smokestack is neither practical nor necessary, says Gassman. Rather, business and tech leaders need to figure out what types of data users need to share, then "bring it together either by bridging silo A and silo B, or by bringing specific data within both silos into a classic data warehouse."

Maryellen Abreu, director of global technical support at vacuum maker iRobot, agrees. "Certainly you don't have to reinvent a whole infrastructure, but rather take relevant data you want to analyze," she says. In the case of customer data, individuals' privacy is also a serious concern, she adds. IRobot's recently completed voice of the customer platform provides a wealth of information about each vacuum cleaner unit, but not the customer's name, email address or any other personal information.

Fitting the pieces together

BI vendors have been fielding self-service BI tools that enable end users to create their own reports and views of data, once it's collected from various information silos. However, they've been slower to work with unstructured data, particularly social media data, which typically needs to be structured and scored for sentiment before it is useful, says Forrester's Kisker.

This leaves IT staffs with the job of cobbling together various vendors' tools and platforms. Take General Mills' VOC platform, for example.

Social media data-service provider Collective Intellect pulls relevant customer feedback data from some 13.5 million Web sites and sends it to Clarabridge's Enterprise text analytics service, which adds structure such as sentiment scoring and brand categorization, and delivers it to General Mills' BusinessObjects-based "reporting universe for social media," says Hagen. Clarabridge also processes unstructured consumer verbatims collected by General Mills' contact centers.

Hagen's team is now working with BusinessObjects to fill in some of the gaps in the VOC platform, particularly when it comes to integration. For example, end users can view customer feedback from social media and contact centers side by side on different BusinessObjects dashboards, but "the two can't be combined at detailed levels," he notes. This is mainly because of the nature of social media: Contact center people can draw out specific information from the customer through questions," but the same process cannot happen with a Twitter or blog post, "so the data can't be as well defined or parsed."

The group is also wrestling with how to consolidate and share information across groups. "BusinessObjects has very robust reporting capabilities, but if you want to create a report about, say, six verbatim conversations that illustrate a particular point, you pretty much have to cut and paste into PowerPoint."

Consumer Services will shortly deploy a BusinessObjects upgrade that allows end users to pull data and generate quick reports on a variety of devices, including smartphones and tablets, Hagen notes. They'll also be able to email the results in a form that a colleague can not only view but manipulate.

Meanwhile, the technical team is "whittling down the range of data" so categorizing and merging it is easier to do, says Bell. For example, "we pull only a small subset of product data out of the SAP system" to refresh the Consumer Services database on a monthly basis. "We don't need to know about raw materials involved in plants, just packaged products for consumer use."

"Voice of the customer means putting information in the ears of people that want to hear it," says Hagen. "But everybody wants to hear something different, so you need a lot of different tools and reporting mechanisms to make it happen."

Help in the cloud

Like General Mills, companies of all sizes have been turning to SaaS and cloud providers to integrate information silos, particularly on the social media front.

BI vendors are starting to extend their platforms into the social media realm, although on a limited basis, particularly when it comes to sentiment analysis. They are also introducing self-service BI services. Other vendors including Tibco, Tableau, Microstrategy and Qlitech led the way, but industry leaders IBM Cognos, SAP BusinessObjects and Oracle, have recently caught up, according to Forresters Kisker.

SaaS offerings such as Tibcos Silver Spotfire enable end users to make the reports that they generate accessible to others, not just to read, but to work on. Because such environments are still part of the BI system, IT can still keep an eye on things, adding popular reports to the standard portfolio, and identifying reports and queries that cause performance problems, Kisker says.

"Certainly you don't have to reinvent a whole infrastructure, but rather take relevant data you want to analyze," says Maryellen Abreu, director of global technical support at vacuum maker iRobot.

IRobot signed up with Rightnow's (now Oracle's) Cloud Monitor service, because it provides "a 360-degree view" of customer intelligence from all channels -- from surveys, emails and phone calls and social media such as Facebook, Twitter, and YouTube, Abreu says. The company monitors and responds to negative opinions on the Web, but does not, as yet, structure and load social media data into its back-end database. That's a future project, Abreu says.

"Rightnow is the front end for our CRM system, but a lot of our business is retail, so we have to collect data from other sources such as the Web, distributors and retailers," says Abreu. "The challenge has been understanding what database fields are important to us: what information [about devices and products] is important to telling the whole story at any one time."

The system is now used by almost every department. Marketing uses the data to track what's happening to a newly released product in the field, week to week. Engineering and quality assurance can enter a returned unit's serial number to call up its full history, including customer complaints, returns and black box data. Quality assurance and marketing can compile complaints across multiple units, to track the incidence of a particular problem. Call centers, once proprietary islands, have been using the platform for about three years, and "call volume has steadily decreased over that time, meaning fewer complaints and problems, Abreu says, although she declined to share the numbers.

Business and IT leaders often see the threat rather than the potential benefits of end users' independent data gathering, particularly when it involves dubious sources such as social media and small cloud-service providers, Forrester's Kisker says. However, he adds, "You can't stop the train."

Elisabeth Horwitt, a freelance reporter and former Computerworld senior editor, is based in Waban, Mass. Contact her at ehorwitt@verizon.net.

Read more about BI and analytics in Computerworld's BI and Analytics Topic Center.

This story, "Busting down the info silos" was originally published by Computerworld.

Copyright © 2012 IDG Communications, Inc.

How to choose a low-code development platform