Confusion between BI and CPM

The term BI originated around 1997. Software industry pundits like to think that every software category with a new acronym is leading edge, but about ten years earlier, vendors had created products with similar functionality and called them Executive Information Systems. Since then, the technology has changed but both the business need and the end user experience are still the same.

In the current software lexicon, Business Intelligence is often confused with CPM. BI is better known than CPM, which is a term that originated around 2001 to encapsulate many analytical software applications including not only aspects of BI but also financial applications such as planning, reporting and consolidation.

Different industry thought leaders started changing the definitions of these terms so that over time BI and CPM were regarded as different software categories that interrelated and later some analysts considered CPM to be a component of BI (this view is especially prevalent in Europe). This all means that discussions about BI and CPM can be very confusing.

Prophix Software views Corporate Performance Management and Business Intelligence as being different categories of software. This differentiation may at first sight appear to be unimportant, but the many products available have different costs and capabilities. An understanding of the market for ‘data viewing‘ software helps an organization choose a solution that is appropriate for its need and that will be a worthwhile investment.

One of the biggest misconceptions is that, if two products are capable of displaying the same data in the same format, then these products are in the same category of software. This is not necessarily the case; in terms of the successful uptake of an application used to display data, the way a user navigates to see the data he or she needs is as important, if not more so, as the actual data on the screen.

Both BI and CPM software can be used to view information using a user interface that includes charts and graphs. The terms can be confused because some of the components of CPM software, especially software that is used to display strategic information such as Key Performance Indicators, can display data in a way that is similar to the functionality of ‘true BI’ products. This common ‘data viewing’ functionality gives rise to obvious questions such as What is the difference between CPM and BI? and Which of them does my company need?

What is CPM?

Prophix follows the ‘formal’ definition, which states that CPM consists of the following five types of software application:

1. Budgeting, Planning and Forecasting software
2. Software used for Financial, Statutory and Management Reporting
3. Applications used for formal Financial Consolidation
4. Software used for Profitability Modelling and Optimization
5. Strategy Management software

Most of these are clearly distinct from BI software,but when strategic metrics and Key PerformanceIndicators (KPIs) are displayed by a CPM application the end result can look similar to BI.

What is BI?

Business Intelligence is one of the software industry acronyms that has multiple definitions depending on which vendor is doing the defining. However, software products that claim to be in the BI space are usually designed to give users access to data in an unstructured interface. To understand the difference between BI and CPM, it is important to realize that the user experience defines the nature of a ‘data viewing’ application.

The user experience

This is often called the application’s Graphical User Interface (GUI). As far as GUIs for data viewing applications are concerned, there is a continuum from unstructured (where the software is highly flexible and gives users the ability to modify greatly what they see) to structured (where users view screens that are pre-defined and cannot be changed to any great extent). Both of these approaches have their pros and cons.

Unstructured products are true Business Intelligence products that have functionality built in to enable users to create their own queries and navigate in their own way to find the data they need. This type of software is likely to include components such as:

• Accessing data over the Internet with an easy-to-use interface.
• Displaying data both numerically and graphically.
• Assuming a structure for data as a basis for data navigation and so that it can be automatically subtotalled.
• Drilling down not only on numerically displayed data but also on graphically displayed data.
• Pivoting displays so that rows and columns in a table can be swapped.
• Sorting the information being displayed and using this as a basis for selecting data.
• Saving views of data so they can be displayed later.

There are many BI products of this type. If there is an existing suitable data source, they usually require minimal implementation and enable users to view information immediately. However, there are risks associated with these products; users may need training in order to derive the full benefits from a product of this type and so ease of use is very important. Other considerations include the difficulty of creating a suitable data source (see later in this white paper) and the ease and quality of printed output; some products that are web-based have limited printing capabilities are really designed to be online viewers of information.

Prophix’s Ad Hoc Analysis functionality provides a numerical interface to OLAP data that is as good or better than most unstructured BI products. With Prophix, users can drill down on data, select the data to be viewed and also access analytical tools to compare data, for example to compare actual with plan. Users with appropriate security privileges can also use this BI style interface to enter data and, with Prophix’s SmartSwap feature, make adjustments to imported data.

Unstructured BI is becoming a commodity, which is reflected by the proliferation of such products and by the fact that Microsoft is offering functionality of this type in Microsoft Office (with Excel 2010) and in Microsoft SharePoint. For a company wishing to experiment with this type of unstructured BI, Excel is an excellent and relatively cost effective alternative.

Structured applications for viewing data usually address a specific need. ‘Structured’ means that not only are users limited to the functionality built into the solution but they also benefit from being guided by the solution to see what is important (i.e. what the organization wants them to see). They are often marketed as performing functions such as ‘ Scorecards’ and thus are considered CPM applications and not part of BI. However, because they display data graphically they are often erroneously called Business Intelligence. It is important to understand this distinction.

These products are often associated with one or more management paradigms such as Balanced Scorecard or Six Sigma and are used for display of more strategic data, usually called Key Performance Indicators (KPIs). By definition, a structured solution limits users to viewing certain data. The big question is What data?, and to answer this question usually involves choosing KPIs that are the most relevant metrics for users to view. Therefore, as well as simply displaying data (including non-numeric data, such as a detailed explanation of why a KPI is important), most of these structured applications also do a lot more. This can include methodologies for the selection and definition of a Company’s strategic KPIs, defining the relationships between KPIs and enabling algorithms to be defined for reporting on aggregated KPIs.

Other examples of off-the-shelf structured applications are dashboards (which simply display KPIs using devices such as gauges, as on a car dashboard) and scorecards. Scorecards are essentially variance reporting where variances are color coded based on rules that need to be pre-defined and is sometimes called Red and Green (RAG) analysis. Scorecarding can also include exception reporting, where exceptional variances are identified and displayed. Historically, these products have also been expensive to buy and implement, but the market is changing and products are becoming more affordable. Both dashboarding and scorecarding tend to bridge the gap between structured and unstructured BI, but the need to pre-define which data appears as gauges or what the rules are for color coding means that some structure is implied.

When implementing a strategic application, it may not be obvious what data to make accessible to users. To help choose appropriate KPIs, many large companies use the services of an outside consultant (who may have experience on a vertical market or in the capabilities of a specific software product) to help decide which strategic, forward-looking KPIs are most relevant to the organization; for a mid-sized company, this can be a major expense. The choice of KPI will naturally change over time (once one organizational problem is fixed, it may no longer be relevant) and so senior management needs to constantly revisit which KPIs are important. The risk is that if a system displays the wrong information then the application becomes irrelevant, people stop using it, and confidence in management declines.

Structured data viewing applications can be expensive. The costs associated with any multi-user software solution consist of acquisition costs (for software licenses and implementation services) and ongoing maintenance (principally services). Structured systems tend to be significantly more expensive than unstructured BI in both these areas. Because of this, a structured solution requires a high level of commitment, especially for mid-sized companies where resources may be limited. Structured solutions can be either off-the-shelf or custom built. The former has historically been sold mainly to large companies (primarily because of the strategic nature of these applications) and this type of software may be expensive to buy and implement.

Custom built solutions are obviously expensive to implement. With a custom built system, anything is possible but many mid-sized companies find the costs prohibitive. When embarking on a custom solution, many companies neglect to consider the ongoing maintenance costs; if application functionality, operating systems or technology change there can be significant future implementation expenses. Custom solutions can be built using a number of tools (such as Microsoft SQL Server Reporting Services) that are bringing down the costs of this type of application.

Structured data viewing applications are often assumed to be strategic in nature, but this is not necessarily the case—they can be operational or strategic. If a structured application is regarded as operational rather than strategic, the choice of what to display may be more obvious. For example, a company can build a structured BI application that displays production information in a shop floor environment. This information will change daily or hourly and is used to communicate to staff the progress that is being made with operational tasks.

One problem with strategic CPM applications is that the data is sometimes not very granular. Financial data or metrics such as headcounts are often available only monthly; users will naturally use the system only when they know new data is available and if this is just once a month, there can be problems with user acceptance. Embarking on a strategic project usually requires commitment from an executive sponsor who realizes the associated value. To be successful, it needs a long-term commitment from the company so that it becomes part of the corporate culture; often, if the executive sponsor moves to a different job, the BI project either is axed immediately or it dies a slow death.

[Download PDF for charts]

This chart summarizes the types of BI products available on the market categorized by type of user experience. There is no market for unstructured custom-built solutions; they are typically too expensive for companies to develop. Off-the-shelf structured products have historically been the most costly to acquire. Some structured solutions allow users to easily drill down to more detailed data using an unstructured BI interface.

Where does the data come from?

The user experience decides what kind of software product a company wants to buy for navigating to and displaying information. Just as important is the source of the data being displayed.

Most data displayed in both BI and CPM applications is internal to the organization that owns the system. This is usually because it is difficult to source timely, relevant external data. We are not talking here about news feeds or other information that is freely available over the Internet; it is usually more cost effective to let users access these sources directly. But there may be external information that is of limited availability and is of value to users; for example, information sold by an industry association that can be freely distributed within the purchasing company. Usually, however, most information is internal.

The data that is displayed in a data viewing application depends on the nature of that application. For an unstructured BI system, users are usually looking at numerical data that comes from a company’s computer systems with a minimal amount of manipulation or calculation. For a highly structured strategic CPM application, data may need to be collected and/or manipulated before people see it. For example, if a KPI is ‘Customer Service calls per Headcount,’ customer service call data may need to be collected from customer service managers while headcount may need to be read from a personnel system and manually adjusted before the KPI can be calculated.

Strategic KPIs are often not readily available electronically and require a system to collect the data from users throughout the organization. Prophix is an excellent application for collecting data from users on a regular basis. Prophix’s functionality for collecting data includes workflow, automatic email reminders, commentary, approval processes, and data entry spreadsheets distributed by email.

ETL and data warehouses

When a company comes to implement Business Intelligence, a common problem is the reliability of the data available. There is a whole segment of the software industry devoted to the task of extracting data from existing systems, transforming it to make it more accurate or consistent (e.g. making ‘Cal.’, ‘Cal’, ‘Calif.’, ‘California’ and ‘CA’ all become ‘CA’) and loading it into a database for eventual display. These are called ETL tools (for Extract, Transform and Load) and are sold mainly to larger companies that have many legacy systems. Mid-sized companies that use Enterprise Resource Planning (ERP) systems often have less of a need for ETL tools since the ERP takes care of storing data in a consistent format.

However, even if a company has an ERP, it may be desirable to create a staging area to store data that is to be consumed by a BI system or to be freely available to many users. This is because directly accessing real-time production data can cause performance issues with those production systems; it would not be a good idea if every time the CEO looked at sales data the invoicing system grinds to a halt.

Larger companies started doing this in the 1970s and created Data Warehouses, which are corporate data stores of information; one of their uses can be as staging areas for BI applications. They were often very big, very complicated and very expensive. Later, the term ‘Data Mart’ became fashionable for smaller data stores that do the same for departmental data. However, depending on a company’s size, these are often too expensive—creating a data warehouse or data mart is not a onetime exercise; it will demand ongoing investment as new applications are implemented and the business changes.

Many mid-market companies use Prophix to create models containing data for BI applications. Prophix includes many of the features found in ETL tools, but with a user interface that is easy for non-IT professionals. Prophix will calculate metrics (e.g. revenue/headcount), can be used as a mechanism for collecting data that is not available electronically, and can create OLAP models quickly and easily. Prophix does not claim to be an ETL product or a data warehouse tool, but one of Prophix’s major strengths is that it can be used by a finance professional to implement business-focused BI applications with minimal need of IT expertise or involvement, skills that are often in short supply in smaller companies.

Where is the data stored?

Most software vendors will specify or recommend a database technology. It is important to understand the alternatives.

Unstructured ‘true BI’ systems usually display larger data volumes than CPM applications. Therefore, good database retrieval performance is essential for the success of a BI application. The data being displayed in a data viewing application usually comes from a database server. Some software vendors sell systems that, for performance reasons, maintain all data in memory, either on a server or in individual users’ workstations. Currently, these architectures tend to be quicker than disk-based systems. However, solid state disk drives are starting to replace physical disks and this trend means that memory-based systems are losing their competitive edge.

Databases used for BI and CPM can be either relational or OLAP. Relational databases are excellent for transaction applications such as accounting systems while OLAP databases are excellent for fast aggregation and retrieval of numerical information. Most people are acquainted with the tabular nature of relational databases and OLAP systems are less well known, though, with the recent popularity of Microsoft’s OLAP component (Microsoft SQL Server Analysis Services), the benefits of an OLAP approach are becoming more widely understood.

OLAP databases are nothing new; they were previously called Multidimensional Databases and have been used in the Enterprise software market since the 1970s. Prophix Software has had experience of OLAP technology since the early 1980s and supports the use of OLAP databases for planning and reporting solutions. Prophix has chosen Microsoft SQL Server Analysis Services (MSAS) as the OLAP database used in its software.

Generally:

• OLAP databases offer faster and more flexible reporting than relational databases
• OLAP databases include the capability of defining more sophisticated built-in calculations that are part of the database instead of the program querying the database
• Applications that use relational databases typically demand more understanding of data structures by end users who wish to report on data

For large-scale BI applications, relational solutions usually have worse performance than OLAP systems of similar size and scope. However, for smaller applications, relational databases can be adequate and there is even an industry buzzword: Relational OLAP (ROLAP) for software that makes a Relational database look like an OLAP database.