The percentage of a typical portfolio of alliances that should, in the ordinary course of events, be expected to fail is a fascinating subject. Well fascinating to those, like us, who analyse alliance portfolios and those alliance managers whose bonuses depend upon convincing senior management that, as a result of their skills, the company has a lower than expected alliance failure rate.
So what is the expected failure of rate of the typical portfolio of alliances? Well various figures ranging from 30% to 70% are frequently thrown around in articles and presentations of the subject. But where do these figures come from and why the astonishing disparity between 30% and 70%? Are these figures based on empirical research or are they estimates based on somebody’s subjective experience in the area? In search of the answer to these questions we thought that we would conduct some meta-analysis (well, that’s what our academic colleagues call it anyway) and go off and find the source of these figures.
So what did we find? Well, we found that most of the estimates originate in research conducted by consultants. For example, the following estimated failure rates were quoted by the Corporate Strategy Board in a report titled Institutionalizing Alliance Capabilities (August 2000):
|McKinsey & Company||1993||33%|
|The Darden School||1996||60%|
|Coopers & Lybrand||1998||50%|
|The Lared Company||2000||60%|
So that’s where the 30-70% range came from. But, can we rely upon these surveys to produce a reasonable expected failure rate of around 50%? Probably not. These surveys were conducted in a very different economic environment than that which exists today. The business environment has changed out of all recognition since 1993 and today, the discipline of alliance management has changed beyond all recognition and information technology has fundamentally altered the way in which relationships are managed both within and without the enterprise. Given those facts, it seems difficult to believe that those surveys provide a reliable guide to a reasonable expected failure rate today. So does more current data exist? Apparently not. Since 2000, other than the bi-annual surveys conducted by Silico Research and IBM in the biopharma space, we seem to have encountered a dearth of other surveys exploring this area. Of course we may be overlooking research so we would appreciate any links to more recent empirical data.