• United States



by Vawn Himmelsbach

Beef Up Your BI Strategy

Jul 28, 201010 mins
Big DataBuild AutomationCIO

Last year CIOs faced budget cuts, project delays and increased demand for services, and this year IT budgets have – for the most part – remained flat. For CIOs, it’s a challenging time, and one that requires them to move from cost-cutting measures to providing strategic value.

Last year CIOs faced budget cuts, project delays and increased demand for services, and this year IT budgets have – for the most part – remained flat. For CIOs, it’s a challenging time, and one that requires them to move from cost-cutting measures to providing strategic value.Agile BI Development: Meet User Demand and Beat IT Complexity

BI Vendors Finally Embrace the Spreadsheet

One strategy is increasing the use of information through business intelligence, according to Gartner’s CIO report, “Leading in Times of Transition: The 2010 CIO Agenda.” We’re seeing big shifts as CIOs reshape their infrastructure, moving from traditional IT to IS lite (taking away certain applications and outsourcing them) to lean IT (where they run a core set of applications) and adding value through innovation.

“This is the transition CIOs are thinking about and it’s where BI sits,” said Bill Hostmann, research vice-president with Gartner, during the Information Builders Summit 2010 in Orlando. BI started out with reporting and dashboards, then moved to ad hoc queries and data mining – now it’s moving into predictive analytics.

For CIOs, the value is in moving users toward a self-service model of information delivery, which can lower the total cost of ownership, drive business costs down and provide data to users on a timely basis.

Evolution, not revolution

Cambridge Memorial Hospital (CMH) is one organization that’s taking it slow. BI, after all, is an evolution, not a revolution. “The amount of data is astounding,” said Ed Norwich, CIO of Cambridge Memorial Hospital. “But the amount of immediately usable decision-making data is quite restricted at the present time, so we want to determine which pieces to provide.”

CMH, a large community hospital in Ontario, rolled out an IBI WebFocus BI platform in order to develop an emergency room tracking board for patient flow and care delivery in the ER. Prior to this, they relied on a whiteboard to record patient information and manage loads – which clearly didn’t integrate with the hospital’s electronic management system.

The tracking board provides an overview of each patient, including basic information and whether lab tests have been completed. This is integrated with the Meditech hospital information system so clinical support departments can view the patient load in ER and estimate the resulting workload. It’s also allowed CMH to monitor patient wait times in the ER.

“It’s readily viewable and easily understandable and it tells us when people have been routed through the lab,” said Norwich. “We have all the critical elements we need to make a decision about patient flow, about support, about the management of cases.”

But while they want to apply BI to other parts of the hospital, they want to get it right before growing their capabilities and knowledge base. While BI is seen as a means to engage staff, that’s often easier said than done. The biggest challenge is a cultural one – getting managers engaged so they actually use the data. At CMH, managers are sent a monthly report so they can start using self-service options, “but it’s like pulling teeth out of elephants,” said Norwich.

Today’s demand for metrics cannot be effectively met using Excel spreadsheets, he said, but to make BI work there has to be recognition by hospital executives that this is the way to move forward and they’ll do what’s necessary to make it happen. “It’s not an easy journey. It’s like going into the wilderness without a GPS,” he said. “Everybody needs to know what the objective is, how to get there and what their part is. Everybody is responsible for being a leader.” The skepticism leading up to the implementation of the tracking board was unreal, he said, but now they can’t live without it.

From reporting to predicting

While BI is an evolution, the holy grail is predictive analytics – not just understanding what happened and why, but predicting what will happen in the future to make more informed decisions.

The Richmond Police Department in Virginia, for example, is using BI based on the idea that criminal behaviour follows identifiable patterns and can be used to predict criminal acts. The department uses previous crime statistics (from their 911 system, computer-assisted dispatch and records management system that tracks crime) and external factors (such as neighbourhood, time of day and even moon phases) to estimate when and where crimes might occur. “Ask any fire department – they hate working on the full moon,” said Stephen Hollifield, CIO of the Richmond Police Department. “We don’t know why, but there’s more crime.”

The department has more than 50 million records of data, but they couldn’t dig into it, so they were looking for a way to integrate that data to find hidden patterns and ultimately put an officer at the right street corner at the right time. Hollifield spearheaded an initiative that integrated the capabilities of IBI WebFocus with SPSS predictive analytics and ESRI dynamic geographic information system mapping, which allows them to look at interactions between present and past data such as arrest records, motive and type of crime at a particular location based on day, time, weather and coincidence of public events.

This data is analyzed on a continuous basis, so when officers head out on a shift, they can see predictions of crime hot spots, essentially giving “green” officers the same kind of intuitive ability that a seasoned veteran would have. The department maps crimes in six different categories, so an officer can see all break and enters, for example, in a particular sector between noon and four p.m. “We don’t want them sitting in front of a computer,” said Hollifield.

But does this actually work? Using predictive analytics, the department has seen some rather dramatic results: the city went from being the 5th most violent city in the U.S. to the 99th. They’ve seen a 246 per cent increase in weapons seizures. On New Year’s Eve, they saw a 49 per cent reduction in random gunfire incidents. Typically, 911 gets flooded with calls on New Year’s Eve, and the department used to flood the city with cops; now they’re able to use BI to predict credible threats and, as a result, save a significant amount of money in overtime costs. BI can also be used for calling in additional support when needed, and to justify the cost, allowing them to move from a reactive to a proactive position.

BI for social media

What impact do blogs, tweets and other social networking tools have on customer sentiment? More companies want to know what kind of coverage they’re getting online, or they may be questioning how well social media actually works. An article in the Wall Street Journal, for example, may be weighed differently than comments from a blogger. And comments from a blogger may be weighed differently, depending on whether that blogger gets 10 hits or 100,000 hits. They also want to know what’s being said about their competitors.

But companies are struggling to get a handle on the blogosphere, said Jim Davis, senior vice-president and chief marketing officer with SAS, at SAS Global Forum in Seattle. So we’re seeing the emergence of social network analysis and sentiment analysis, which identify “influencers” in social media.

Social media is, essentially, a new data source – an unstructured one – and companies want to know how to drive business decisions out of that data. Automated sentiment engines don’t necessarily do the job, since not all social media tools are equal. Twitter, for example, is spiky in nature, while a blog provides more opinionated views and feedback.

SAS introduced Social Media Analytics earlier this year, an offering the company says will analyze online conversations to drive insight and performance; it’s available as a managed solution hosted by SAS Solutions OnDemand. The company acquired text analytics vendor Teragram in 2008, so they can now provide unstructured data analysis around internal and external sources of information, coupled with the ability to do predictive forecasting and correlation. SAS uses advanced data mining to come up with an “influence score,” so companies can see who’s talking about their brand and how influential they are.

“This stuff is here to stay and it can have a major impact on your business – and we can’t ignore it,” said Davis. “Small groups that you may never have heard of can have a major influence on your business.”

The Marriott International is one SAS customer that sees value in connecting the growth of social media to its business. They want to understand where their business is coming from, so they’re using social media analytics to find out how well their brand is received in the blogosphere, who’s influencing that conversation and how that sentiment is divided up by source, whether it’s blogs, forums, news or reviews. An influence tab provides information on top Twitter influencers and top Twitter hashtags, for example, and a “phrase cloud” pulls together comments on hot topics such as “sustainability.” They can then drill down through that data to look at specific comments.

Dealing with change

This, however, is part of a much more comprehensive BI strategy at the Marriott, where SAS is used for data management across the company’s portfolio, from sales and distribution to marketing to customer relationship management to revenue management. The company has 550,000 hotel rooms to fill each night; in addition, they’re looking to open new hotels and start new brands around the world. Being able to offer the right price through the right distribution channel is critical, said Mike Keppler, senior vice-president of global sales, marketing and revenue management systems with the Marriott.

The challenge, however, was that even though they had a number of credible metrics that could help run the business, team members were not ready to accept those. “They didn’t understand it, and they weren’t ready to make decisions on those metrics, so it’s important to get the leadership team wrapped around those metrics that help you make decisions,” he said. “It doesn’t matter if you’re in charge or part of the team, it’s going to be challenging.”

He recommends starting with a business and IT partnership, so when you walk into a room of thought leaders you can’t tell who’s wearing the IT hat. And talk about the most important business objectives, such as where you’re going to place your resources or where you’re going to source that talent.

Regardless of the stage they’re at with BI, companies need to retool as we come out of this downturn. “It’s not necessarily about best practices, it’s about new practices,” said Davis. “Innovation involves risk and disruption. It’s not about doing what you’re doing today better, but taking it to the next step.” But in this economy, companies are being careful, and too much optimization is seen as a challenge to long-term viability. “Are these at odds? Is optimization the enemy of innovation?” he said. “Maybe, but it doesn’t have to be. A balance can be struck.”