This paper tries to start a discussion about new ways and methods for calculating the potential return on investment (ROI) for case-based reasoning (CBR) projects and systems in use. It is motivated by the fact that as soon as CBR is regarded as an established technology, the return on investment question is one of the first serious hurdles that have to be taken. From my experience, prospective CBR users are not quick in either answering this question or in arguing why this question is not adequate – which means that this topic needs more attention, arguments and support for how to deal with it. 1 How is this Topic Related to Business Processes? The topic of this paper does not correspond exactly to one of the three workshop categories but it has relations to all three of them. The question “Will it pay off to build and use this CBR system?” should be answered at least roughly at the very beginning of any CBR project and it can (and should) be repeated after each prototype development cycle in order to incorporate new insights gained during the last prototyping cycle. The calculation of the return on investment (ROI) is – besides the system development and system integration costs – to a great deal influenced by the efforts for case knowledge acquisition, for the case model and case base maintenance and last not least by organisational impacts. 2 Why do we need New Ways of Looking at this Topic? With the internet age, the classical methods for calculating assets are becoming somewhat old-fashioned. “Systematic management of intellectual capital creates growth in shareholder value. This is accomplished, among other things, through the continuous recycling and creative utilisation of shared knowledge and experience.” – Leif Edvinsson, Skandia AFS [Bach et al. 99, page 25] There are first examples for firms like Skandia that have started to integrate knowledge management activities into their daily business processes and even publish a knowledge balance-sheet with their annual report. Österle [Bach et al. 99] states that “knowledge creates shareholder value” and examples for this hypothesis are e.g. the market quotations for firms like Yahoo or Amazon or eBay and a lot of other internet start-ups that went public. All of these firms • do not possess significant assets of the classical kind like ground, buildings, machines, materials etc. and • do not earn money but produce rather big losses – at least during their early years. The only thing these firms really “possess” are innovative ideas and business models that come together with sufficient knowledge, abilities and power for gaining a substantial market share within the new markets they have helped to create. From these internet start-up cases I think we can see that classical ROI methods are not suitable for grasping or explaining what is happening here. 3 Why can’t we just Borrow or Inherit Appropriate Approaches from Knowledge Management Projects ? About one year ago, I thought that all these KM projects started in big consulting firms will eventually lead to accepted models for estimating the monetary value of reusable knowledge and that the CBR community should be able to borrow or inherit from such models. But till today, there is – in spite of activities within accounting standards committees no result of this kind visible and therefore I am beginning to doubt whether “wait and see and adapt what the KM community has agreed upon” will work for solving our problem – at least not on a short-term basis. Maybe these KM projects are still in a stage where the upper management thinks that these projects are of strategic importance and this means “don’t ask about the ROI yet” but CBR has never been in such a position and therefore has to learn rather quickly how to deal successfully with ROI questions. So this paper is a pleading for starting to think about and for developing approaches that show how the knowledge contained in case bases can be given a reasonable monetary value. 4 What can we Propose Regarding the Value of CBR systems? 4.1 The Classical Approach The classical approach essentially says that the costs of an application can be determined in the following way: • Calculate the efforts for building the CBR system and putting it into use • Add the efforts for using and maintaining the CBR system for about 3 to 5 years Its benefit can be determined in the following way: • Calculate how much time and money is saved by re-using a suitable case retrieved from the case base • Multiply this result by the number of expected case base usages and the probability to find a suitable case in the case base for the same period of time (about 3 to 5 years) as above. This classical method is appropriate in situations where there are very many users around and not so very many cases so that every case has a rather high re-use rate – at least on the average. This method is less appropriate in situations where e.g. the number of users is limited in the beginning and it is unclear how quickly the number of users and with them the re-use rate will grow. 4.2 A non-classical Approach My problem with the classical approach is that the acquisition of a case makes money just fade away – which is not the case when I buy a computer or build a house. People argue that I can touch a computer or a house and that I can try to sell it if I don’t want to use it any longer – and that this is not true for “virtual goods”. This is exactly the point where I don’t agree. My view is that nobody would invest money in documenting a solution if (s)he would not be rather sure that this investment will pay off – and therefore the investment just changes its form to a knowledge asset but does not fade away. Knowledge assets have an actual value that can even surmount the initial effort for their acquisition, especially when their re-use rate is high and their contents is attractive for other firms using the same equipment and having the same sort of problems to solve. I admit that we are far away from a case base market because most firms explicitly don’t want to share their knowledge with their competitors. But not only competitors might be interested, also the developers or manufacturers of the faulty equipment could profit from the experience contained in the cases, user interfaces or manuals could be improved with this material etc. and all such improvements generate value from the contents of a case base. Therefore I propose an alternative case evaluation model along the following lines: • The initial value of an acquired case is equal to the effort invested in its acquisition. • This value rises when the case is frequently re-used – the higher the re-use rate the higher the value of the case. • This value goes down when the case is re-used very rarely or not at all. This way of evaluating cases has the consequence that a case base administrator is responsible for deciding which cases are worth being documented with how much effort to invest. Furthermore (s)he should explicitly aim at gaining more users and improving the CBR application in order to preserve and rise the value of the case base. For case bases that could be sold to other parties an additional component comes into play: • The actual value of a case base consists of its internal value plus its market value – even if nobody thinks about sharing the case base because it represents a competitive advantage.
