Usamos cookies y otras tecnologias similares (Cookies) para mejorar su experiencia y proporcionarle contenido y anuncios relevantes para usted. Al utilizar nuestro sitio web, acepta el uso de Cookies. Puede cambiar su configuracion en cualquier momento. Politica de cookies.

The Best Architecture for Business Intelligence

Originally published noviembre 13, 2007

Supporting enterprise performance management – the ability to gain business intelligence insight across your entire organization – requires a strong technology foundation. Only by establishing an information architecture that provides consistent data models and integrated tools can companies be assured of achieving their goals in performance management.

This is not merely a technical issue; it has strong ramifications from the standpoint of business. Currently, most companies have difficulties integrating data from multiple sources, such as enterprise resource planning (ERP) applications, line-of-business (LOB) applications, existing data warehouses, and business intelligence applications. But with a data model that provides consistent and thus highly reliable information, companies can do more than improve decision making. They can also increase the speed of decision making, monitor business drivers linked to strategic objectives, create new business opportunities, measure operations performance against forecasts, and increase efficiency and data quality.

To achieve these objectives, however, companies must start by constructing a data model that takes into account multiple needs – human, technological, and even regulatory. A structurally sound data model must be well integrated so that companies can verify the source and veracity of any given piece of data. It must establish common definitions and a policy for using them. And it must accommodate users’ application needs.

Before building the optimal data model, however, companies must understand their current state and how it stifles their goals of data integrity and consistency. Unfortunately, most companies’ business intelligence initiatives are run at the line-of-business level, using department-specific point solutions, rather than solutions driven from an overarching perspective.

The result: data isolated in silos, originating from different sourcing mechanisms and housed in applications using different data models. These range from Excel spreadsheets to specialized databases. Because of this, there is no way for C-level executives to gain an accurate, aggregate view of the company’s current performance or trends in sales, costs, returns, or other factors. For instance, one company’s vice president of sales may define revenues one way and the CFO in the same company define them another way. Because of their different perspectives, the CEO gets a different answer to the same question. As a result, executives are frequently forced to manually massage data into consolidated reports that dilute the accuracy of the numbers. This not only takes time, but it also diminishes the reliability of the data.

To combat this, companies must construct an enterprise data model based on the following underlying enabling capabilities:

Establish Common Definitions

To a company’s shipping department, a customer is represented by an address; to a company’s finance department, a customer is represented by an outstanding balance. For enterprise performance management (EPM), these views must have a common definition. Similarly, revenue as defined by sales must be defined the same as revenue defined by accounting. This gives a consistent view across the entire company.

This requirement of common definitions extends deeper, however, into both the business and technical metadata. Business metadata describes key performance indicators (KPIs); technical metadata describes where the data came from. Within your data model, you need to be able to verify any particular piece of data by tracing it back to its origin. As a result, both IT and the business side must work together to create a consistent definition that accommodates these multiple elements of an EPM data model.

Map Your Processes

Once companies have established their definitions, they need to map their business processes – that is, track the end result of business activities back to their original data sources and repositories. There are two reasons for this. First, the company must ensure that it can correlate the elements contributing to a strategic business target with its operational foundations. For instance, if a company’s strategic goal is to reduce product returns, it needs to understand where all the information collected about product returns resides. In this regard, a company must certainly avoid silos of information and be cross-functional, understanding how to accommodate information from both the supply chain and marketing. It must also accommodate the inclusion of data from nontraditional sources, such as customer satisfaction surveys.

Second, given the increasing emphasis on auditing in light of regulations such as Sarbanes-Oxley, companies must be able to prove the veracity of the information they report to stockholders. Understanding the source of their results simplifies their auditing efforts.

Accommodate User Access

Though this facet has a human element to it, it is fundamentally a technology-enablement issue. A key stumbling block to EPM deployments is lack of user adoption, especially if users are forced to use unfamiliar applications. Depending on what the users are familiar with – Excel spreadsheets, a business intelligence application, or even a Web-based portal – the data model must accommodate the way that users work, and integrate those applications with the enterprise-wide data structure so that data is controlled consistently at a central level.

Consider Application and Data Integration

A company’s data model must be flexible enough to accommodate the integration of its existing applications. This is no small feat in itself, but it also must accommodate new dimensions of information. For instance, if a company targets a new type of customer, it needs to track results of that foray. The data model must accommodate that addition and technically be able to propagate it throughout the front-end applications being used. Few companies can add a new column to 150 spreadsheets in 25 departments fast enough to accommodate its decision-making process.

Finally, companies need to consider their ability to integrate new applications and information into their new corporate data model. Any new application that a company deploys must offer APIs or development environments that allow its IT team to easily integrate that application’s information into the corporate data model.

Building such a holistic data model may sound complex – and indeed it is. It is a difficult transition for companies to make, but one that can only be done as a company-wide project. Deploying an enterprise performance management program requires a C-level perspective and a significant investment commitment of several years.

The payoff, however, is a clarity that can have positive ramifications for years to come. With the alignment that comes from a consistent data model, a company can discern trends sooner and react to them faster. EPM, based on a strong data-model foundation, is a key way to inject agility into an organization.

SOURCE: The Best Architecture for Business Intelligence

  • Markus SprengerMarkus Sprenger

    Markus is a BI Global Solutions Director and Avanade's primary business intelligence (BI) expert. He defines and directs the implementation of Avanade's solution strategy relative to Microsoft's BI products and alliances, including the creation of intellectual property (IP) and reusable implementation assets that accelerate customer deployment.

    Markus leads a team of solution architects who work closely with the Microsoft Office Business Applications product group. His team influences the Microsoft product road map through the escalation of technical learnings, challenges and feedback from Avanade customers and the global BI community. Markus joined Avanade in 2005, and has more than 10 years of experience in designing and delivering Microsoft-based BI solutions. Prior to joining Avanade, he owned a business intelligence consulting company in Germany and worked in the management of a BI-focused ISV.

    Editor's note: More financial articles, resources, news and events are available in the BeyeNETWORK's Financial Services Channel, led by Markus Sprenger. Be sure to visit today!

Recent articles by Markus Sprenger

Related Stories



Want to post a comment? Login or become a member today!

Be the first to comment!