TOP TOC 
Agile . since Jun 23 . Index . DOCs TOP TOC
In today's management jargon, best practice is everything. Best practice represents a safe harbor in a stormy sea of too much hype and too many problems. Best practice means that someone big and successful has used something that has been blessed by someone else. Best practice means that I don't have to defend my decision against outside consultants or inside critics.
Best practice is, therefore, the Good Housekeeping Seal of Approval. Unfortunately, because it is such a great term, best practice is overused and, in most contexts, has lost much of its original meaning. Moreover, best practice is hard to quantify. Which is really software best practice, CMM or agile development? Both sides have data and proof to support their definition of software development as a best practice.
Agile . since Jun 23 . Index . DOCs TOP TOC
What's worse is that in the 21st century, best practice may not be good enough or fast enough. I am a big fan of Clayton Christensen and his idea of disruptive technologies. Christensen became interested in this issue when he was doing research for his Ph.D. at Harvard Business School. He was intrigued by the fact that he saw numerous business examples of established, very well-run companies lose out to startup companies in areas where older companies had once dominated.
In particular, Christensen was fascinated by the fact that none of the companies that were big in the 1980s manufacturing 5.25-inch disk drives were able to segue their dominance into the even bigger market for 3.5-inch drives. And as he studied other marketplaces, Christensen found similar cases in other areas such as discount retail (Wal-Mart versus Sears), software (Microsoft versus IBM) and steel (Nucor versus US Steel).
What Christensen concluded was that these companies were not poorly managed; for the most part, they were very well managed -- for a given environment. In each of these markets, the dominant firms did what was expected of them: they worked to keep growing and to maintain the profit margins, and they listened closely to their best (largest) customers. In a word, they represented the best practice for their time and for their industry. And, ironically, being the best practice led to their downfall.
As I interpret Christensen, these dominant, best practice companies became trapped, naturally enough, by their own success. As they grew in size, they had to take over more and more of their market to maintain the growth rate. And they had to do this while maintaining their profitability. This made them natural targets for disruptive technology startups that, although not ready for prime time, found niche (low-cost, specialty) markets where they could grow rapidly, learn how to make money with lower margins, and then expand into the major market. Christensen found this process to be repeated over and over again in market after market.
Startup organizations that embraced disruptive technologies learned to prosper with less overhead and less cost. In time, when they overtook the bigger, slower rivals, they became the best practice (and ultimately subject to some other disruptive technology).
So what are the conclusions that can be drawn from all of this with respect to best practice? Well, if an organization -- no matter how well-managed that organization -- is not careful, it can be overtaken and replaced by another practice that it can't understand or integrate into its operation. I call those practices based on disruptive technologies "next practice."
Increasingly, organizations in the 21st century will need to be focusing on what the next practices in their industry, market, or specialty will be. In the software world, I believe that next practice has to do with knowledge versus effort. Let me illustrate this by pointing to Figure 4, which has two dimensions. The horizontal dimension is effort or, if effort is too hard to track, amount of documentation. The vertical dimension is value or knowledge.
 Figure 4 -- Next practice quadrants (knowledge versus effort). |
It seems to me that this figure represents the future of software better than the CMM levels. Here we have four quadrants instead of five levels.
| | | |