Last week, the Sloan Management Review published a study by Andy King and Baljir Baatartogtokh that re-examined the success record of Clay Christensen’s theory of disruption. They took 77 claims/predictions of disruption and asked experts to evaluate whether what happened to those companies was consistent with the theory of disruptive innovation. They found scant support.
It is worth reviewing why. King and Baatartogtokh take a very literal view of the theory of disruption finding four elements that Christensen explicitly identified as comprising the theory. Those elements are:
- Incumbents are improving along a trajectory of innovation (53/77)
- The pace of sustaining innovation overshoots customer needs (17/77)
- Incumbents have the capacity to respond but fail to exploit it (47/77)
- Incumbents flounder as a result of disruption (48/77)
The numbers in the side are the number of the cases that possessed the element while only 7/77 possessed all four. It is on this basis that the authors conclude that the theory of disruption is not that useful.
In my reading, however, I think we need to be somewhat more circumspect. In particular, there is an issue of distinguishing between Christensen and his co-authors’ over-enthusiasm for seeing disruption all over the place as opposed to finding ways of testing the theory and advocating for its usefulness.
For instance, element 1 is a selection device. The theory of disruption applies when firms are being well managed and are responsive to their current customers. If in 24 of the 77 cases this wasn’t happening then we shouldn’t be looking at those cases to test the theory of disruption in the first place. Thus, they should be discarded. It is only conditional on them being in the 53 appropriate cases that we should examine the theory.
Element 2 is also problematic. To be sure, Christensen emphasised it. But, in reality, it is a side-show. You don’t need ‘over-supply’ to have disruption arise. Instead, you could be supplying your current customers optimally and still get into trouble according to the theory. Thus, while it is useful to examine element 2, I am not sure we need to require it to obtain usefulness out of the theory. In any case, what does it mean? How can you ever tell if you are getting over-supplied even in cases where all consumers are identical, which, of course, they are not. In my mind, this part of disruption has always been a sideshow and we should ignore it if we are really interested in the phenomenon of successful firm failure.
That brings us to element 3. This is the issue at the heart of disruption — basically, where the incumbents are disadvantaged relative to entrants. Of course, they are disadvantaged not because they are protected from competition but instead because they are doing well and that gives rise to issues in responsiveness. Thus, King and Baatartogtokh are correct to identify cases listed by Christensen where incumbents could not respond because of, say, regulatory constraints as not being cases that are consistent with disruption. In other situations, threats came from so far away that it was not really fare to say the incumbents could have responded. Christensen would disagree I suspect but this is really a situation where the theory becomes murky and so, strictly speaking, such cases cannot be really classified as being in the potentially disruptive camp. I say this because disruption really has its power in challenging managers who have the capacity to respond. Sometimes that does not occur. That said, I often hear that BlackBerry could never have responded to the iPhone threat because Apple was too good. I don’t believe that BlackBerry managers should get off that easily, however.
The final element is more compelling because it gets to the heart of the notion that disruption disrupts incumbents. It appears that in a good deal of cases this didn’t occur but King and Baatartogtokh do not really delve down enough to say why. In addition, floundering is a very weak test of disruption theory. Disruption theory is a theory of firm failure. Now while failure may not always mean exit (BlackBerry is a case in point) it also doesn’t mean stumbling or having trouble. It means a devastating blow from which a firm does not recover. I could not tell what fraction of the firms in the sample had been disrupted by this metric — a metric that Christensen seems to use.
While it is a tough exercise to use case counts to investigate disruption theory, it can illuminate the applicability of the theory in the form exposited by Christensen. However, for this to occur we would want to do the following.
- Examine only the 53 cases that satisfy element 1 as these are nominally successful
- Ignore performance on element 2 as it is a side-issue
- Tighten up the definition of failure to be stronger than just floundering
- See which of the 53 cases satisfied element 4
That should give us a clearer test of disruption theory itself although the general point that the Christensen cases do not give full support as a whole to his own theory still stands.