Understanding the concept of learning curves is critical in the deployment of process improvements. Misunderstanding learning curves can result in the abandonment of a project during the prototype phase because the new methodology under performs the established work processes.
This phenomenon was evident when Tiger Woods basically took a year off his game to learn a new swing. At nearly every point throughout that transitional year one could have argued that he took an enormous step backwards with his “new” approach. However, by the end of that year it was evident that he had successfully scaled a new plateau. John Henry and his quixotic race against the steam engine may also be an apt metaphor.
Any job has an associated learning curve. The learning curve is a model, ranging from a rule of thumb to a sophisticated mathematical representation, which predicts how long it will take, on average, for a new employee to become proficient at that job. “Proficient” is a key term in this definition as it captures both accuracy and speed.
It may take only a few minutes to memorize the steps associated with a particular task, it may take days, weeks or months to turn out 50 widgets an hour that are 99% defect free.
Training is typically used to show an employee how to perform a task, but activity on the job, under the watchful corrective eyes of an experienced coach, is needed to scale the learning curve at an appropriate rate. As in piano lessons, perfect practice makes perfect. Perfect practice is always slow because almost anything can be done wrong quickly. By forcing a slow enough pace to perform the task perfectly during the early going it then becomes possible to add speed without sacrificing accuracy.
The numbers element of a codified learning curve is essential in gaining speed at a rate that will build throughput without losing accuracy. Going too fast too early is self-defeating, but staying perfectly slow for too long obviously sacrifices productivity. People rarely understand how fast they can comfortably and accurately perform a task until this is demonstrated by experienced hands.
I once spent a summer, at a large apartment development, stapling felt underlayment to plywood floors in preparation for pouring concrete. I was reasonably dedicated, thought I was pretty fast, and whacked away at it for a few weeks doing a room every half-hour or so. The site foreman wandered by one day and watched me finish a unit. After a few minutes he picked up my roll of felt, asked for the stapler, and told me to follow him into the next room.
By the time I got into the next room he was walking into the room beyond that. He was stapling a whole room faster than I was walking between rooms. I knew exactly how to do the task, it had never occurred to me that it could be done that fast. Getting up to speed wasn’t that much harder, but it required a different mindset and recognition of the possibility.
Whether recognized or not the learning curve is always there. The impact of the learning curve on a process improvement or technology enhancement introduces a latency factor into the ultimate performance of the new system.
Experienced employees transferred from the legacy system to the new system experience a reset on some portion of their learning curve and their performance will lag for a period. Worse, if the new employees are forced to work the new system at their existing productivity rate then they miss the opportunity to engage in “perfect practice” and may be significantly delayed in regaining and surpassing their old performance.
Too many genuinely improved systems fail at this point if the project team has not budgeted for sufficient training. Because the experienced employees cannot afford to fall behind in their current work long enough to learn the new system, they reject the proposed improvement outright.
By the same token there are “improvements” which are marginally worse than the legacy or manual processes they were designed to replace. Without a well-defined learning curve established, in advance, it is impossible to determine whether the system is performing better or worse during the transitional period; and how long that transition should require.
Some consideration of the impact of learning curves is an important part of the change management process on any improvement effort. Taking this critical factor into account can eliminate many false starts.
Copyright © 2007, Lotus Pond Media
Steven Grant is a former customer service executive from American Express with over 25 yearsdevoted in Fortune 500 companies analyzing, improving and delivering on enhanced customer experiences.Share your experiences and suggestions on improving the customer experience at http://www.customerresearchcenter.com or email Mr. Grant at scgrant@customerresearchcenter.com