Q: How can the quality department prove to top management that the department adds value and not cost? What are the best things we can do to make sure management realizes how important we are to the organization?
A: Being able to draw a straight line between what you do for your organization and the value it adds to the organization’s bottom line is an important ability, especially when a department or group is not directly linked to a profit center.
Companies are in business to make money and to return value to their stakeholders. Our business unit received a Baldrige site visit in the early 1990s. At the kick-off meeting, our business unit manager welcomed the examiners with the following greeting: "Welcome to our division, where we make money and grow the business."
Blue Ocean Strategy1 is an excellent book about business development strategy. One of the key concepts covered involves driving customer satisfaction higher while reducing costs. I have long held this as a personal definition of what quality is all about.
So, you have customer satisfaction, expense reduction and money making. But how do they all come together? This is where the secret lies. Individuals in the quality arena must be able to connect the dots between what they do and the reasons a company is in business.
The first thing a quality department needs to be able to do is tell the story of how quality helps the company make money. I’m a big believer in the supply chain, and I think quality represents multiple links in that chain.
To that end, help your management understand those links and how you help the supply chain operate more efficiently or profitably. An individual’s place in the supply chain will determine how that explanation goes. When you do this, make it all about the business and not about quality.
Use metrics to show how you are helping the company make money. Look at your part of the budget, figure out what your cost drivers are and go after them. If possible, leverage third-party benchmarking to help you establish this.
You will find that many organizations struggle with leveraging data the way people in the quality field can. That’s why quality professionals are perfectly equipped to help organizations make sense of their data. Target an internal organization and see if you can help it make a decision based on the data. In fact, a great question to ask them is: When was the last time it used data to make a business decision?
Adopt a mindset of becoming an internal consultant to the business. Again, with the varied backgrounds of many quality professionals, many practitioners are adept at playing the role of a business solutions consultant. Show your management it can preserve capital by letting you run some of these cost-saving projects and proving you can keep money inside the organization. Never forget: It’s about the business, not about quality.
It should not be a big surprise that everyone in the business world has an agenda. Many of these agendas are simply to maintain the status quo. Some departments may avoid the publication of a more thorough set of metrics, fearing they will show how overstaffed their organization actually is.
Try to learn about these agendas and learn how to debunk them. After you gather your data, make a pitch to your management. More than likely, they will listen intently.
Remember, quality is everyone’s job, not just the quality department’s job. Quality is achieved by everyone in the company, not just those that have quality in their job titles. Help people, especially those in quality, realize this. As you learn how to tell that story, it will help you elevate quality to a new level at which it truly belongs.
Director, continuous improvement
Certified quality engineer
- Kim, W. Chan and Renee Mauborgne,
Ocean Strategy, Harvard Business School Press, 2005.
Q: I have a question about proper control charting that I’ll illustrate via an example.
Part A is in a family of functionally similar cast-ductile iron parts differing only in size. The stability of the casting process over time is a point of dispute. The time unit basis chosen is a calendar month.
We have 16 months of data for nonconforming parts per month. Currently, normalizing stability as a function of volume for part A is not possible. Monthly counts of the total number of part A processed are not known for the purpose of this question. Stratification of process stability as a function of size or other non-time variables will be undertaken later.
One measure of the casting process quality is soundness, which is defined as the absence of voids on surfaces machined in the raw casting process. If part A has a void on a machined surface, the part is classified as a nonconforming part. Quantity of voids per part isn’t counted and isn’t relevant for this question.
The quantity of part A machined each month is variable, ranges from 24 to 90 and is not known absolutely for any of the 16 months. The findings are as follows:
- Month 1 = 1
- Month 2 = 4
- Month 3 = 14
- Month 4 = 9
- Month 5 = 2
- Month 6 = 5
- Month 7 = 7
- Month 8 = 7
- Month 9 = 6
- Month 10 = 0
- Month 11 = 2
- Month 12 = 10
- Month 13 = 11
- Month 14 = 22
- Month 15 = 21
Can these observations of the quantity of nonconforming parts per month be control charted legitimately to discuss the stability of the casting process as a function of time? If yes, which control chart is appropriate? If not, then why not?
A: Count data occur in many different real-world situations. Counting manufacturing defects, typos in a book and accidents at a certain intersection all have the characteristic of count data.
Counts are typically expressed per some unit of observation or interval. For example, manufacturing defects may be expressed per units inspected, typographical errors per page of a book and accidents per thousand cars.
Control charts based on the Poisson distribution, specifically, the c-chart or u-chart, may be used to understand the stability of data based on counts. But for the question at hand, the problem is that the unit of observation varies monthly and is not exactly known.
For example, if you graph the count data by month, you’ll see a spike in nonconforming parts in months 14 and 15. But is the spike due to an increase in defects because more components were produced, or has the rate of failure per component changed?
To illustrate, suppose the likelihood of having a failure on a component is estimated to be 0.3. If 10 units are produced, you would expect to see three nonconformances. If you double your production to 20, then you would see six nonconformances. If you didn’t have that unit of observation, you might erroneously conclude that your process stability has changed when, in fact, it hasn’t.
Even though you can chart the counts and see a pattern, the interpretation of the results will be ambiguous and misleading. It’s never enough to just throw numbers in a software package without understanding the underlying assumptions of the data and the research question at hand.
Packaging quality specialist