So we have had great response post on “A Comprehensive Approach to BPM.” The whitepaper from that blog post can in the whitepaper section of www.processgenie.com In the previous post, to Mitch Gooze of the Customer Manufacturing Group ( http://www.customermfg.com/ and http://valueaccelerationblog.com/ ) touched upon Ichiro Ishikawa’s “Seven Tools of Quality.” Some of you asked if he could briefly review them. At ProcessGenie…your wish is our command…Enjoy!
We all talk about the Seven Tools of Quality. We all know they are comprised of:
Cause and Effect Diagram
Althought all are fundamental tools of process. Let’s walk through each briefly.
Histogram. A histogram is a bar chart used to present visual data from a process over a period of time. Using a visualization of the data a person is more likely to see and understand issues relating to the process which are not as obvious from a bunch of data elements.
For the purposes of business process management a histogram is a useful tool to help make better decisions. As with all visualizations of data you must be careful to create a visualization that helps make better decisions as opposed to supporting the decision you want to have made. (We have all seen graphs that have been scaled to make it look like either huge changes or no changes have occurred based on how the graph is scaled. Histograms are no exception.)
How many bars in your bar chart?
To create a useful histogram you need at least 50 data elements and 100+ is better to make useful decisions. The number of “groups/bars” that you review is based on the number of data elements. The rule of thumb to determine how many columns or bars to have is to take the square root of the total number of data elements and round to the nearest whole number. For example if you have 150 data elements, the square root of 150 is 12.25, so we round to 12 columns or bars.
If you wanted to create a histogram of website hits over time, then the time interval of interest would be based on the number of web site hits you get. If you are low volume website you may get 100 hits/day, so you may want to view a histogram by hour, or maybe even by day of the week. If you are a high volume website that gets thousands of hits per hour, you may want to view a histogram by 5 minute intervals.
You look at the shape of your histogram to determine where is it is centered and what is its shape (though some processes are naturally skewed). How variable is it. If you are running a call center and your midpoint is 100 calls per period and it ranges from 2 to 200 that is obviously much more variable than if it varies from 180 to 220. Is variability a problem and if so, how can you correct it? For example, in call centers they will often monitor “hold time.” However, the acceptable hold time from the company’s perspective for a customer wanting to place an order may be much different than the hold time they find acceptable for a customer wanting to complain. Either way, is the process meeting the stated requirements.
Like most good process monitors, you would expect your histogram to follow a 6-Sigma distribution such that virtually nothing fell outside +/- 3 sigma from the mean.
Cause and Effect Diagrams (also know as Fishbone Charts because of what they look like) are a graphic representation of the possible causes of a particular problem or condition in a process. The purpose of the Diagram is to help find the root cause problem. Using the fishbone approach the team
looks at possible causes in increasing levels of detail. The tool is effective for a team to use to brainstorm possible causes of non-performance or under-performance of a process or of particular failures within a
process. For example, if leads (prospects) are periodically entered incorrectly into the system, what might be causing that to happen? The Cause and Effect Diagram focuses the team on causes not symptoms and keeps the conversation focused on the problem and not personal agendas.
Alternatively to brainstorming, which is a very common approach to creating these diagrams, data from Check Sheets can be used to provide useful insights into the actual causes.
Check Sheets. A Check Sheet is a simple and easy method to collect, accumulate or count data from a process of interest. While it does not seem sophisticated (which it isn’t) and it is easy to master, with one critical issue, these do not mitigate the effectiveness of Check Sheets. Check Sheets are easy to use and
understand, the key is to agree in advance on what you are observing and counting. All you are doing with a Check Sheet is collecting, accumulating or counting data from a process over time to observe or learn what the data tells you.
For that reason, while the task (counting) is easy, it is useless if you do not agree on the definition of what is being counted. If you are counting website hits per hour, define what you mean by a hit. For example is that unique hits or all hits? If you are counting trade show visitors, how long do they have to be in the booth to be considered a “visitor?” If they just drop their business card in the bowl for a chance to “win” does that constitute a visitor or do they have to talk with someone. While it is useful if the definition passes a reasonableness test, it is understanding, acceptance and agreement on the definition that is critical. Once you agree on what to count, then you just count it and tabulate it on the Check Sheet. Check Sheet forms should be simple, clear and easy to use.
You can use the data collected from the Check Sheet directly by observation. For example, collecting errors on order entry forms can show that a particular portion of the form generates most of the errors. Alternatively, you may use the data collected to create a Pareto Chart or Histogram for more complete or complex analysis.
Pareto Chart. A Pareto Chart is a bar chart in which the measured values are arranged in descending order, and it is usually accompanied by a line graph over-laid on the bar chart showing the cumulative category percentages from right to left. Like other of Ishikawa’s tools, the Pareto Chart provides a visual interpretation of data.
The Pareto Chart takes advantage of what has become known as Pareto’s Law, also commonly referred to as the 80/20 rule. In many cases when you graph the causes of “problems” with a process using a Pareto Chart you will find that the 80/20 law is alive and well and you can quickly see where to focus your efforts in process improvement to gain the greatest impact.
A simple use of a Pareto Chart could be to graph causes of customer defection or lost opportunities. When we have done that for our customers, the 80/20 rule has proven to hold true.
Control Charts. Control Charts are used to provide a visual indication of variance in a process. All processes have variances. The question is how much variance is acceptable. The 6-sigma folks use that metric as their guideline, but many processes need a much lower variation (if airplanes landed safely only at a 6-sigma rate, nobody would fly), and some can allow more variation. Though sometimes we accept too much variation based on an assumption that better control is either not possible or too expensive. Japanese car manufacturers created cars with much tighter tolerances than their U.S. counterparts and actually saved money in the process. To use Control Charts effectively, you must decide how much variation is acceptable in the process of interest. That sets your upper and lower control limits. By charting variations in the process over time you can determine if your process is in control or not. If not, then the key is are the variations due to common causes which are inherent in the process and if they produce unacceptable performance require a process redesign; or special causes which must be eliminated to keep the process under control.
Scatter Diagrams. Scatter Diagrams are a visual tool used to determine if two variables are related. While it is a visual tool, done correctly it is also a valid statistical tool to determine the strength of correlation (if any) between two variables. Scatter Diagrams are created using a simple two dimensional graph, with one variable on the x-axis and one on the y-axis. They are often used as a follow-up to a Cause and Effect Diagram to determine if there is actually a valid relationship between two variables as postulated during the Cause and Effect brainstorming session. For example, you could test whether errors in order entry were correlated to the training given to the order entry person. Scatter Diagrams do not predict cause and effect; they only show the strength of any possible correlation between two events. It is also possible that a Scatter Diagram will show a negative correlation between two variables. For example, there have been studies done in the past that showed that some ads actually reduced business for a company.
Flow Charts. Flow Charts are also called Process Maps. They are the most commonly used tool in business process improvement. Simply put they are a graphic representation of the process of interest. The key to Flow Charts or Process Maps is to create them at a level of detail necessary and sufficient to allow improvement to the process. No more and no less. The first step in Flow Charting is to create a picture of the “as is” process. It is very tempting to try to change the process when charting it because too many processes, once visualized, are quite silly in some of the steps they include. Resist the temptation and map the “as is” process before attempting to improve it and creating a “should be” or “to be” process. There are many graphic representations of Flow Charts or Process Maps that are useful, so pick the one that is appropriate for your needs.
But don’t just use Flow Charts. Using all Ishikawa tools is a great way to begin working to improve your existing processes.
PS Download the Pareto Charts whitepaper at www.processgenie.com