As a service delivery company, most of the objectives our customers use to measure performance are based on averages: Average Time to Resolve, Average Up-Time, and so on. Even in engagements where our delivery teams are “All Green”, meaning we meet or exceed all Service Level Agreements, customer dissatisfaction may persist.
This is largely due to the fact that our end users do not consistently experience the averageservice but the extremes that make up the totality of the service provided. Think of it like this, you buy a sauna and want an average temperature of 85oC, but the temperature fluctuates between 50oC and 150oC degrees while mainly heating around 70oC degrees.
Would this satisfy you? No, it would not and this type of variation does not satisfy our customers either.
This is where Statistical Process Control (SPC) can help deliver highly satisfying customer service while reducing operational costs.
Statistical Process Control (SPC) is a process that uses statistical tools and measures to proactively monitor process performance over time. It is accomplished by visually observing the common and potentially special variations of the process and correcting each type of variation with the proper corrective strategy. Its use was pioneered by noted scientists Walter Shewhart and W. Edwards Deming in the early and mid 20th century. Statistical process control is credited in helping the United States manufacturing industry achieve great success in World War II and rebuilding Japanese manufacturing after the war. While Statistical Process Control has its roots in manufacturing, there is now a great deal of evidence of its success in service environments.
Statistical process control closely relates to Continuous Improvement by focusing on process visualization, identifying the causes of variation and applying the appropriate mitigations to the cause and not the result. Variation ultimately increases service delivery times and the probability of waste resulting from rework.
Consider the sauna example; while the temperature varies between uncomfortable temperatures (normal temperature 65-85 centigrade degree), the user inevitably spends energy readjusting the thermostat to achieve the desired comfort level in the desired time frame. The net result is customer dissatisfaction and wasted energy.
Statistical process control improves customer satisfaction and reduces waste by reducing service delivery time variation and the probability of rework related to the variation. The normal objective for service processes is to deliver a service within an expected average time frame. The unwritten expectation from customers is that the average service delivery time does not have wild or non-standard deviations from the expected average. Most service delivery environments lack the tools and measures to account for the variation between the times a service is delivered.
Control Charts are the most prevalent tools and Descriptive Statistics such as Standard Deviation are the most common measures used in Statistical process control . These tools and measures empower Statistical process control through their ability to examine a process and illuminate the volume, severity, and type of variation being experienced.
Control Charts visually display process variation that may negatively affect the quality of service delivery. Viewing process variation will not tell you the root cause of a problem, but it will tell you when a problem occurred and the type of variation that produced the problem. Control Charts will tell you if your process variation is Common (probable based on historical performance) or Special (improbable based on historical performance). Knowing the specific type of variation tells you if there is a problem with your process (Common) or if you are experiencing an external influence (Special). Control Charts are also effective as they can illuminate developing trends that may produce a service failure.
Managed service success is often measured on how well a team restores service after an outage. However, these teams should be deemed successful by their ability to identify problems or failures before they occur or early enough to mitigate service limiting outages. Here is where Control Charts can prove beneficial. Control Charts visually display a process performance as it begins to set a trend or shift. Trends and shifts are indicative of process changes that are outside of statistical probability and driven by external influences on the process. They manifest within the normal or common range of variation, but are not common variation. The influences may not immediately cause a process to fail or miss a Service Level Agreement (SLA), but will definitely indicate a problem that has the potential to cause a failure. This visual indication can forewarn operational teams of impending process failures that could result in service incidents allowing them to mitigate and avoid the failure.
Hypothesis Tests are statistical tests used to validate or reject the assumption a process has changed. This benefits operation managers by providing advance notice if process mitigation is necessary or not. Similar to control charts, they can tell you if a process performance has changed using data samples a change can be determined prior to the end of a reporting cycle.
A Service Manager responsible for data center capacity management applied Statistical process control to his country’s data center environment. The Control Charts deployed monitored the ongoing daily capacity used in the environment under management. Over time a clear trend was established that allowed the capacity manager to accurately predict the point in time when a specific application environment would fail due to a shortage of capacity. Using this advance notice, requisitions for new server capacity were submitted and approved along with the necessary changes. The environment was upgraded shortly before the capacity breach occurred, thus preventing a Priority 1 outage for a critical application to a top customer.
SPC does require nominal initial investments in developing solid measurement criteria and statistical training. But, the results are well documented and the benefits far outweigh the investment. When properly deployed, SPC can provide a tactical advantage over quality practices that rely solely on post problem correction.
If you want to learn more about statistical process control, check out our in-house SPC training using Minitab® Statistical Software