Most of the large software providers have some business intelligence related offerings. Microsoft offers the Microsoft Dynamics package. IBM is the largest provider of such software. And there are many other smaller options. These software packages share several features:
- they are providing graphics for managers with overviews of key performance indicators
- they aid in selecting KPI's
- they include options for database queries
What they don't do is more important when we are considering how such applications could be improved. They don't tell you whether your key performance indicators are well-aligned with the overall strategy of your organization. They don't tell you which data are most relevant in order to make good decisions. Software can be made in more intelligent ways to support such tasks.
Let us first take the issue of alignment with strategy. Most of these systems include the concept of balanced scorecards in some form or the other. The concept of a balanced scorecard is well-known among managers because this is now standard material at any business school. The balanced scorecard is supposed to be a snapshot of an organizations current "operating state" and was developed at Harvard in the 1990's by Prof. Kaplan and co-workers (Kaplan, 2001) . The idea is as follows. Any company must be very clear on what its purpose is in order to succeed in a harsh and competitive environment - in other words, you need to know what you want and what you are doing to survive in a free market. When this "company mission statement" is clear and agreed upon, a strategy must be laid out on how the company is practically going to achieve these goals. A very common approach in determining the strategy is to select some business areas to focus on - and then to formulate performance measures within each category. This makes the strategy more manageable and is also the foundation of a scorecard. A typical scorecard looks at the company from a few different angles:
- the shareholder view,
- internal business processes,
- customer view,
- innovation and learning view.
These are four categories suggested by Kaplan and co-workers - and they seem to fit many organizations quite well, at least those which strive to make a profit for the owners, a.k.a. reguler firms. Within each of these categories, a set of performance measures is selected. These key performance indicators need to provide good information on how the organization is doing in that field and should be closely watched. The problem is just that selecting them can be quite difficult; there are no generally accepted methods to do that and the selection is often done based on intuition and experience. Much better results could have been obtained with mathematical models and optimization methods - formal decision processes that can be implemented in software and that can be updated when there are large changes in the organizations operating environment.
In order to formalize KPI selection, a mathematical model of the company is needed. Such models can be built based on transaction data, HR data, customer interviews and so on - activities that generate a large business database. Textual information must be translated into numerical forms by appropriate coding setups; methods well-known in statistics. From this large set of data, it is possible to do principal component analysis and end up with a relatively low-dimensional linear model of the data: a model predicting how outcomes change when you change of the aims are applied in your business system. This model, together with a clear definition of the vision, or strategy, in mathematical terms can be used to automatically compute a set of functions of business outcomes that are measured on a regular basis, that are such, that when they are kept close to some pre-determined values the company will perform well and according to strategy. Such an approach would create KPI's that are as closley aligned with the strategy as possible with the available business information: a huge improvement over the common situation today where the strategic alignment of the preferred KPI sets are questionable at best.
If such a formal procedure is going to be implemented together with a balanced scorecard system, a performance measure must be established for each of the four categories above (or, generally, the categories in use in the managment system). This could yield a better scorecard implementation.
The day will surely come when model based KPI designs arrive - a situation longed for by auditing firms - imagine the profits that are possible if you can produce a continuous auditing system: a highly specialized and craved-for software which reduced the need for expensive-to-hire specialist accountants. Continuous auditing is definitely coming, and principal component analysis, database analysis and optimization theory will surely play a big part in this very exciting future of business intelligence systems. Maybe these systems will actually be intelligent, some day. What is needed to reach this situation? Research in applied mathematics, database conversions and a lot of software development.
Therefore, if you are considering buying a business intelligence system, make sure you are not ending in a vendor lock-in situation with one of the big software houses; the future may come from an entirely unheard-of company: small startups are often the more innovative ones.
Kaplan, R. S. "Transforming the Balanced Scorecard from Performance Measurement to Strategic Management: Part I." Accounting Horizons 15.1 (2001): 87.
- Business intelligence systems are not intelligent
- KPI selection can be automated and aligned with company strategy
- Continuous auditing is possible in the very near future