In 2010, the QFD Institute convened its 22nd Symposium on QFD, in conjunction with the 16th International Symposium on QFD. Having attended and reviewed all the case studies of these conferences, I have begun to assemble of list of "don'ts" which may explain why some companies fail at QFD.
Based on some recent case studies and emails I reviewed, here are some of the common themes. I hope you will find these tips useful in your NPD projects.
"Our QFD training did not address our product..."
Most lean sigma programs equate QFD and House of Quality (only) or a four matrix approach (4-phase model). This is common among programs that are booked-trained, since most non-Japanese language books on QFD only cover this approach.
Historically, this 4-phase model was truncated in the 1980s by U.S. auto parts manufacturers who were looking for a simpler way to imitate the Japanese Quality method, rather than actually understanding the core of the Comprehensive QFD approach, which was developed by Drs. Mizuno and Akao (the founders of QFD).
As one of the few people outside Japan to learn QFD directly from both of these QFD founders, I can assure you that the 4-Phase model was the exception, not the rule. It was based on a 'reliability deployment' study (not even 'quality deployment' study) which was done at Fuji-Xerox to address a specific component problem.
So, for component manufacturers addressing reliability issues, the Classical approach (4-phase model, HoQ centered approach) might be still a reasonable approach — although one should note that those approaches were created specifically for the 1980s business model (does your company still use the 1980's business method?). If build-to-print manufacturing is not your business model, however, being trained in the 4-phase or HoQ approach is sub-optimal. It is especially so for those who make systems and end products, those in the service, processing, or software industries.
In today's resource constrained businesses, the most efficient approach I have found is to custom tailor the QFD process to meet the needs of your product development process, taking into account management style, product, customers, competitors, and other factors.
"How do I use 'house 2' or 'house 3?' "
Continuing the above discussion on the 4-Phase QFD model, note that the House of Quality is just a specific kind of L-matrix, one of the 7 Management and Planning tools. It is just one instance of one tool out of many. Because the matrix is such a powerful tool and/or owning to superficial QFD understanding, some trainers try to make it do too much. (Like the movie, "the matrix is everything.")
There are other, often more efficient tools, that can do the job of the matrix more quickly. Encourage your lean sigma trainer to go beyond book-learned QFD and introduce you Modern QFD tools, such as Blitz QFD®.
"My matrix got so large, we did not have time to finish it..."
The function of the L-matrix in QFD is to show the relationships between two discrete data sets, and then transfer the priorities from one set to the other. Matrices can grow too large when we violate this first function by having more than two discrete data sets.
"When relating the rows and columns in the matrix, we could not easily agree on how much weight to enter into the cell..."
There are numerous ways to improve this. First, agree on a standard format of which data set will go in the rows and which in the columns.
In many Japanese examples, more attention was paid to picturing the flow of data from one matrix to anothe,r and so rows and columns were often switched to make the flow charts easier to follow. Many trainers and books do not realize this shift in the angle when teaching those old approaches. Another issue is the number of levels of weights — generally for a subjective decision, the traditional 3 levels is insufficient to yield an accurate assessment.
" 'Whats' and 'Hows' confuse me..."
Me , too. Originally, the expressions were mnemonic shortcuts created by one of the first QFD learners in the US auto industry in the 1980s. Initially helpful, the terms were later embedded in QFD software applications and books and have been a key source of confusion and misinformation cascades, for both experienced trainers and new students who attempt QFD with proper introduction.
The terms "Whats" and "Hows" originally referred to the rows and columns of a matrix: What was to be delivered in the rows; and How to deploy it further in the columns. In this way, it is not wrong. What made it confusing was that customer needs (the rows of the House of Quality) were confused with "what" the customer wants the product to do (which is really product function) and "how" the product will do it (which is really enabling technology).
A better way to describe the rows of the House of Quality should be "why" the customer wants the product and a better way to describe the columns should be "what" the product should do. In other words, the "what-how" matrix should be actually a later deployment, not in the House of Quality.
"Explain why the Analytic Hierarchy Process (AHP) is more precise..."
Dr. Thomas Saaty, creator of AHP, received the 2007 Akao Prize for his contribution to QFD, and his work details why the eigenvector better models human judgment than other approaches. Here are two key points: People can make better judgments with words than with numbers and when comparing two issues that are close in value. Thus, paired comparisons applied to a hierarchy can produce an eigenvector that closely approximates the relative importance of the items being compared.
The math in the original House of Quality matrix was created when the Japanese workplace was still using an abacus. Although many scholars and book authors still follow the outdated math (which they blindly took from the 1960s-1980s publications), there is no reason why you should follow their mistakes. Such math is no longer valid for today's Six Sigma / Lean Sigma environment, and it is not suitable for today's computer age.
"Why wasn't AHP used in early QFD case studies, if it was better?"
QFD was formalized in the 1960s in Japan and the matrices added in the early 1970s. Supercomputers used to do the eigenvector calculations and were not easily available to quality engineers and personal computers (PCs) did not hit the market until the 1980s. When AHP software for PCs became available, Dr. Akao and his colleagues immediately recommended replacing the ordinal rating scales used in early QFD.
Unfortunately, this was after the U.S. auto industry had begun publishing the old math in books and articles, which were then spread around the world. Those of us who continued to study QFD with Dr. Akao and his colleagues also made the change and this is what we teach today in modern QFD.
"If we use AHP to get customer importance ratings, must we also use AHP for other math calculations in QFD?"
AHP gives more precision than ordinal ratings. If you custom-tailored QFD process and then use matrices to deploy customer needs downstream (most QFDs will not need matrices at the start), take care not to dilute this precision by re-introducing ordinal ratings in the relationship matrix, competitive benchmarking, FMEA, and elsewhere, especially when it is actually easier to employ AHP.
"Lots of terms are used these days, such as voice of customer, customer needs, wants vs. needs, demanded quality, etc. What works best for QFD?"
When using advanced tools such as QFD and AHP, domain expertise is necessary to get accurate prioritizations. Thus, for customers to accurately prioritize customer needs, they need to have deep domain knowledge or subject matter expertise. Customer needs reflect the discrete data set where they are experts.
"How does the gemba help get customer needs?"
There are many sources of raw "voice of customer data" such as interviews, surveys, field reports, etc. Gemba is a source unique to QFD. In the Japanese quality movement, gemba usually refers to our company's manufacturing floor, a source of information about processes and the root cause of failures. This is sufficient for existing product defects, however, the power of QFD lies in assuring the successful quality of future produces BEFORE the production gemba begins, i.e. in design. Thus, we go to the customer's gemba to identify what failures they have in their processes, as well as successes that must be protected during redesign.
"Are some gembas more important than others?"
Gemba visits require resources and therefore we must strategically plan to do the most important visits first. Gemba planning not only requires identifying who to visit, but the ideal conditions to collect the best data. This is serious part of modern Blitz QFD®.
"Who should prioritize the customer needs?"
As described above, for accurate prioritization, domain knowledge is required. Customer needs are best known by -- customers! Not the QFD team. There are actually some books and articles that erroneously recommend the QFD team do the prioritization.
These and many more causes of failure can be avoided by gaining the sound and solid QFD foundation through the QFD Green Belt® and QFD Black Belt® courses. They are offered several times a year as public courses. They can be also arranged as an in-company training (send an inquiry to QFDI). Application white papers and case studies are available through transactions page.
New product development is one of the most valuable business processes a company can have. It keeps the product pipeline flowing with innovative products to enrich the lives and productivity of our customers. It is the engine that continuously drives profit into the organization.
It has been estimated that over 80% of new products fail in the market, but QFD can improve that by focusing our best efforts, first on what matters most to customers. QFD is thus too important to be left to antiquated models borrowed from other industries and other times. Your QFD must be yours, must be fresh, must be done properly in order to assure maximum customer satisfaction with the most efficient effort. Don't you agree your customers deserve no less!
© QFD Institute