For many companies, business analytics works when a handful of very sharp people with complementary skills get together to tackle a project. The results can be rewarding, but often they aren’t translated into an ongoing, repeatable process. Effective companies turn business analytics from a craft, practiced by a few analysts and decision-makers, into an established scientific approach that dramatically contributes day-to-day to the company’s bottom line.

This article shares some successes in taking an enterprise data warehouse and harnessing it to user-driven advanced business analytics through the use of an analytical sandbox. We see analytics as not something that business users beg the IT department to perform for them, but rather something the IT department facilitates so business users can explore, analyze, predict and forecast, and automate key processes.  

Getting Off the Treadmill – For Good

Every sports store can tell you that come August, you should have plenty of soccer balls, cleats, footballs and shoulder pads available. In late winter, the baseball gloves need to come out. In early spring, the swim goggle shelves should be fully stocked. This seasonal stocking doesn’t seem to be something that requires much in the way of analytics -- football is played in the fall, baseball begins in the spring and kids swim in the summer.

But what happens if the YMCA opens a new indoor pool in your area? Or the recreational soccer leagues move up fall and spring play by a week? What happens when a fall baseball league forms? A spike in off-season sales will occur, but you won’t know to fully take advantage of that trend unless you’ve got an incredibly alert manager. Meanwhile, customers who can’t find needed equipment search for it online – and possibly never return to your business.

An advanced analytical approach does two things: It helps business users discover these trends by providing them with up-to-the-minute data in a user-friendly way, and it helps to automate – via alerts and dashboards – certain processes. Business users, for instance, might see a dashboard showing products with the biggest out-of-season sales jumps and, by digging into the data, can see if an automated suggestion to stock more goggles at “Store A” makes sense.

An advanced analytics strategy relying on accurate data flowing quickly to business users can dramatically affect many areas. Here are just a few examples:

  • Analyzing warranty claims can spotlight factory floor problems and help a company fix them before the claims become too costly.
  • Automating the process of drawing up marketing lists allows companies to target the market in cost-effective ways that boost sales and profits.
  • Applying advanced analytics to fraud detection can help insurance companies, manufacturers and government agencies detect a fraudulent claim before the claim is paid.

What’s Required?

There is, of course, a catch. None of this is possible if data can’t be acquired and analyzed in near real time by business users. As we discussed in the first part of this series, data locked in silos or analyzed in isolation from other critical data is of minimal value. Effective analytics can’t be dependent on submitting custom data pull requests to IT or shipping data out to third-party vendors to analyze. The analysts need access to the most recent data available, and they need access to it where it sits (typically in a data warehouse). In addition, the analysts need to be able to store the results of their work back in the database so that they are available for immediate use by other analytic processes or reports. The role of an analytical sandbox is to facilitate these efforts.

Let’s consider the real-life example of a successful sporting goods retailer. The company uses a sandbox approach to assess the value of the company's advertising efforts. Analysts can measure the success of each communication vehicle in a campaign conducted simultaneously with email, catalogs and retail fliers. They can also investigate how each medium interrelates with other media. Being able to perform most of the data preparations without the help of IT – which is simultaneously working on other important data warehouse initiatives – saves time for IT and helps the marketing team gain the benefits of the data that much faster.

In addition, the technology makes the company more agile. With the data warehouse, the company receives each day's sales data by the following morning and can assess conditions and proposed steps – such as launching an email campaign – to mitigate problems and boost sales.

Keeping Your System Tuned

Many people have seen the classic “I Love Lucy” episode where Lucy and her sidekick Ethel attempt to keep up with chocolates streaming down a conveyor belt. You can’t help but laugh as they begin eating the chocolates and hiding the rest. In the analytical world, the “I Love Lucy” episode’s equivalent is the solution that flags so many outliers that a decision-maker can’t keep up. You want the decision-maker and analyst to be able to refine and optimize the process to reduce outliers over time.
A terrific example of this comes from the insurance industry. It is not unusual for large insurers to be working with data sets of tens of millions of rows or more. Solutions promising to use algorithms to detect outliers and alert investigators of suspect claims often create too many false positives, or it takes too long for IT to format the data to run a fraud detection solution against it. Because many contracts require insurers to pay within a certain period of time, investigations are initiated after the health care claims are paid. This model typically yields no more than 5 percent of dollars lost to fraud.

What happens when your data is coordinated and your analytics are highly advanced? One state prevented $14 million in fraudulent Medicaid claims from being paid out and detected an additional $27 million in fraudulent claims that led to indictments. A private insurer detected and saved $11 million in one year when using a finely tuned solution. Investigator productivity climbed 30 percent as activities that once took a day now take minutes to execute. A solution can be tuned to kick out cases to just the right business user. Seeing a spike in gym shoe sales? That will go to the buyer who orders shoes. Is there a jump in questionable, high-cost medical tests? An investigator with appropriate skills in that area will receive the alert.

Those sharp people we mentioned in the first paragraph of this article are still doing what they do best. But they are able to do more of what they do – and do it all in a more timely fashion – through the use of an analytical sandbox embedded within their EDW. The sandbox enables not just faster, more current analysis, but also easier distribution of the results of that analysis to the right people. Most importantly, leveraging an in-database sandbox enables analytics that just aren’t possible without one. Better information getting into the right hands faster – who would argue with the value of that concept?

In part 3 next week, we'll discuss the best analytical functions to automate and why a vendor's promise to "automate everything" is an empty one.

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