Back ] Home ] Up ]


The Half-life/Complexity Matrix


Arthur M. Schneiderman

If you are using a state-of-the-art improvement process, you should be able to close the gap between your current performance and the ultimate process capability at a constant rate given by the process half-life.  The half-life is not the same for all processes.  Instead, it depends on the complexity of the process.  There are two dimensions to complexity: technical and organizational.

Technical complexity is high for new technologies, where part of the learning process is related to understanding and refining that technology.  Over time, as the technology matures and its use becomes more routine and familiar, technical complexity declines.  It is the nature of business evolution that we are constantly dealing with a mix of old and new technologies.

Organizational complexity arises when a process has linkages (intermediate inputs and/or outputs) to processes outside of its boundaries.  These processes may be internal or external (outside suppliers or customers) to the organization.  The linkages may be one-way or two-way, one time or interactive, routine or requiring real-time negotiation.  So processes can run the full organizational gambit, from completely self contained (uni-functional) to cross-functional or cross-organizational.  We would expect the rate of improvement to slow as the cultures, goals and objectives of the various players come into potential conflict.

By sorting the improvement efforts that I studied into the nine bins formed by classifying each in terms of the process's apparent organizational and technical complexity (low, medium or high), I arrived at the following matrix of average half-lives:

For processes of low organizational and technical complexity, I observed a rate of improvement of 50% per month, while for processes at the other extreme, the rate of improvement slowed by more than a factor of 20 to 50% improvement every 22 months.  Note also that organizational complexity had 2 to 3 times the effect of technical complexity in slowing improvement.

Although my determination of complexity was somewhat subjective, I've had little difficulty getting a team of knowledgeable people to agree where a given process falls on this matrix.  In the rare case when consensus fails, the arguments always involve adjacent elements so that a simple average will break the deadlock.

This half-life/complexity matrix can be used in several ways.  As a goal setting tool it allows rational determination of future performance based on the use of state-of-the-art incremental improvement tools and methods.  As a diagnostic tool, it allows a team to benchmark its improvement efforts against best practice for processes of similar complexity.  As a measure of organizational learning, it is easily consolidated from the level of the team, through the business unit to an overall organization-wide metric.  As a specification for an improvement process, it allows potential users the ability to make comparisons with alternative methodologies.  

However, a warning is in order: in practice, many organizations fall well short of these target half-lives.  There is a strong temptation to say the target half-lives are wrong rather than to admit to inadequacies in their improvement methodology.  In the end, the competitive environment determines which of these is correct.  

Note that flattening organizational structure or organizing around processes rather than functions does much to reduce organizational complexity and thereby increase the potential rate of improvement (that is reduce the process half-life).

As a final word, when managers and industrial engineers owned the incremental improvement process, their achievements could be summarized in a similar matrix.  My guess though is that the numbers in their matrix were about ten times higher.  Why?  Because there were too few of them to address the vast number of problems that form the basis of the performance gap.  But organizations learn, and the empowerment of all employees to work on improvement is the key learning of the last half-century.  What's next?  Sharing of best practices through benchmarking and networking, facilitated by the internet, has the potential of shortcutting some of the time consuming steps in the improvement process.  Time will tell. 

back to top

1999-2006, Arthur M. Schneiderman  All Rights Reserved

Last modified: August 13, 2006