1/1/2001 | 8 MINUTE READ

Better Decisions Made Better

Facebook Share Icon LinkedIn Share Icon Twitter Share Icon Share by EMail icon Print Icon

Business intelligence tools excel in exposing the subtle business trends that require immediate attention, instead of just exposing the apparent business hemorrhages that any competent executive can see and repair.


Facebook Share Icon LinkedIn Share Icon Twitter Share Icon Share by EMail icon Print Icon

Business intelligence (BI). Sounds oxymoronic, yet BI is a response to the adage: You can't manage what you don't know. The recent problems at Bridgestone/Firestone give credence to this: having raw data about the company's tires was not enough.

So more and more businesses are turning to BI tools to make sense out of the torrent of transactional data collected by enterprise resource planning (ERP), customer relationship management (CRM), and other data-intensive applications.

This act of “making sense” leads to a better understanding of the strategic matters and trends that affect your business, which leads to more informed decision making.

What's BI?

BI tools fall into that software genre called “decision support.” Sometimes also going under the acronym EBI (enterprise BI), these tools generate short- and long-term views of one's business using the existing data captured by the company’s information systems.

Granted, conventional structured query language (SQL) and reporting tools are supposed to do that, too. However, conventional tools are geared for one-time ad hoc or canned queries and reports, while BI tools are designed for users who want to “follow their bliss” when it comes to analysis. For these people, one question leads to another and another and so on into a continuous series of “what ifs.”

To help such analytic free association, BI tools let users rearrange data to create new correlations, and to discover the new perspectives those engender. “What you're looking for are relationships in multiple dimensions,” explains Dave Van Noord, vice president of Technology and Product Management for BRAIN North America, Inc. (Ann Arbor, MI). “Sometimes you have to really slice and dice the data in different ways to find that relationship.”

Let's take peeling tires for example. Which customers report that their tires are peeling? When were these tires made? What production lines produced these tires? Who was running the lines? Who supplied the raw materials? What was the temperature that day? What was the shipping temperature? How were these tires shipped? Do other tires have the same characteristics?

“There's a lot of value in low-level queries,” continues Van Noord, “but BI helps you move up, to finesse and summarize the data to see trends.”

What's in BI?

BI consists of three basic components:

  • data warehouse, which collects, organizes, and processes data from all sorts of transactional systems for reporting purposes.

    Data warehouses differ from the conventional relational databases used in transactional systems by how the data are structured. In conventional databases, data are normalized and the database is built for high-performance transaction processing. In a data warehouse, the data are de-normalized; the database is built around a subject matter, and all data regarding that subject is piled into that database. This makes the data warehouse more efficient for data access, multidimensional analysis, and reporting. And because the data are now in the warehouse, the transactional system can chug merrily along by itself while the warehouse responds to data analysis-like activities.


    BI system consists
    A BI system consists of three main components: data warehouse, BI reporting tools, and analytical tools. The data warehouse is the keeper of data, which are extracted from external data sources and transactional systems. These data are identified by some sort of metadata file or data model. The analytic tools use this intermediate file to access and then process the raw data. The BI reporting tools work with the analytic tools and the metadata file to display the results of the analysis. Voilà: new relationships, new correlations, and new perspectives to base strategic decisions regarding supplies, production, transportation, sales, finances, and all the other domains in a business.(Source: SAS(Southfield, MI)
  • BI reporting tools, which include on-line analytic processing (OLAP); digital dashboards, those dynamic pages in Microsoft's Outlook 2000; and basic query tools. These tools answer the “what” and the “where,” but don't typically address the “why,” says Thomas Roehm, Automotive Industry manager forSAS (Southfield, MI).

    People usually associate BI with OLAP. According to the Data Warehousing Institute, OLAP tools “provide a visual interface for accessing and navigating through multidimensional data stored in either relational or multidimensional databases. These tools are great for ‘slicing and dicing’ data, but have traditionally been weak in generating complex, formatted reports.”

    This multidimensional quality separates OLAP from conventional reporting tools, such as the ubiquitous spreadsheet. Spreadsheets are two dimensional; they have rows and columns, their intersection being cells. If rows are products and columns are dates, then the cells tell how much of a product were sold on a specific date.

    OLAP tools are multidimensional, like a Rubik's Cube, offers Roehm. Each direction in that cube is a different variable. So if the main item of interest is product, the other dimensions can be date, sales representative, sales by region, product returns, and so on. “Your BI query does not actually hit the raw data; instead it hits this cube of predefined, summarized data,” continues Roehm.


    SNAPshot Enterprise WebSNAPshot Enterprise WebSNAPshot Enterprise Web
    BRAIN's SNAPshot Enterprise Web lets you access and analyze multi-dimensional data. The display in this screenshot, for instance, depicts three dimensions of data: department (number), year, and scrap (quantity). You can drill up and down any of these dimensions, these data sources, as well as filter and sort the data. In this case, as shown in the tool bar with a bunch of folders, you can analyze data by shift date and number, ABC class, scrap reason, and so on. That is, you can create queries and reports based on your business needs versus some prefab query. Plus SNAPshot lets you access, analyze, filter, sort, and display data to your heart's content over the Web, whether across a corporate intranet or an extranet.(Source: BRAIN North America, Inc. (Ann Arbor, MI)
  • Analytical tools, which answer the “why” something is happening based on existing data, include forecasting, statistical analysis, and data mining. Data mining, which is a whole topic in itself, is a technique for selecting, exploring, and modeling large amounts of data to discover previously unknown patterns and correlations, which leads to anticipating future behaviors, events, and consequences.

    What's BI used for?

    Unlike conventional SPC/SQC tools, BI tools combine disparate data from across multiple disciplines to uncover patterns that then become the basis for decision making. “We're analyzing data for everything from production to payables; from scrap, production, and inventory analysis to financial analysis and project reporting. You name it,” says BI user Kelly Knepley, director of Information Systems for the Suspension Component Business Unit of Hayes-Lemmerz International, Inc. (Ferndale, MI).

    Knepley offers two examples where BI shines. First, to isolate a particular casting defect in a foundry, a quality manager might want to know which day and production bay this defect was most prevalent, how many castings were produced, and what percentage of overall scrap the defective casting represents. Second, on the financial side, say a company has a capital project for which it has issued purchase orders (PO) and paid a lot of bills. Now it wants to know exactly what the PO budget was for the project, outstanding committed dollars with no receivables yet, how much cash has been paid out to payables, and what's still owed that hasn't come in yet.

    To answer these sorts of questions, Hayes-Lemmerz, which has been using BRAIN's Trans4M ERP system for about five years, started implementing BRAIN's SNAPshot Enterprise Web in the late spring of 2000.

    SNAPshot is a BI tool that provides operational insight into key business indicators in near real-time. It extracts data from the Trans4M database and then presents those data using a graphical OLAP tool. Its reporting functions work over the Web through corporate intranets and extranets. Because Trans4M is part-number instead of work-order based, SNAPshot lets you drill quite far down the ERP database to focus on profit and loss, on-time product delivery, scrap analysis, and other areas of manufacturing control.

    At SNAPshot’s core is BI/Suite from Hummingbird Ltd. (North York, ON, Canada). The Hummingbird engine transforms raw ERP transactional data into information by providing querying and visualization capabilities. But for that to work there is the ERP data model that BRAIN supplies. This model shows the relationships between the thousands of data elements and data tables that constitute the ERP database.

    Creating this data model is straightforward for BRAIN because it provides the underlying transactional system, Trans4M. But it’s still a lot of work to create that model, so BRAIN is saving its customers from a massive amount of time-consuming chores, such as knowing how to create table joins in SQL and mapping ERP data to the analysis tools. “It’s a very big deal to me to just install a piece of software, load the data model, and have a user who knows about point-and-click to be looking for data in a matter of seconds,” says Knepley.

    For obvious reasons, SNAPshot is focused on the Trans4M ERP system, so its data model is hardly plug-and-play for accessing other data sources, such as external data, other legacy systems, and other data warehouses running off different applications and platforms. But the good news is that other BI tools can access ERP-like data from the SNAPshot “business warehouse” far more easily than from the transactional systems themselves.

    Hayes-Lemmerz’s implementation of SNAPshot is a two-phased project. The first phase is to replace the company’s existing tools for ad-hoc reporting, be they SQL-type queries, Crystal reports, or Microsoft Access, with SNAPshot. This immediately makes it easier for Hayes-Lemmerz to analyze its raw ERP data, as well as to share reports throughout the corporation. The second phase is to implement Hummingbird’s BI/Web, which provides querying, reporting, and OLAP capabilities over the Web and therefore to all of Hayes-Lemmerz’s sites worldwide.

    Getting started in BI?

    BI is not a technology decision; it’s a business decision. Points out Roehm, “You need to clearly define what you’re trying to do with the data and what knowledge you’re trying to gain from that data. Then realize that every business problem is a little different and different tools have different strengths.”

    The BI tool must help you identify business problems quickly. For that, it must have good data access capabilities, be easy to use, and have the functions to lead you to root causes fast. Moreover, the BI implementation must be according to the rules of business, not information technology. That’s key. Muses Roehm, “We hear a lot of stories about ERP implementations that are over budget and provide limited decision-making value. Data warehousing projects can have the same results if the implementation doesn’t have a business focus.”