QFD ANALYSIS: From Customer Needs to Design Specs

New product development (NPD), invention, innovation—call it what you will, but all refer to basically the same discovery-analysis-creation process. Not that many in the product design biz could describe it fully; many work on an intuitive level. One of the primary, formalized methods to help in this process is quality function deployment (QFD), which got a bad name in the 1990s. “It was multi-failure. A lot of practitioners thought it was just chart drawing and changing voices of customers to specifications. It wasn’t,” says Jim Finn, QFD unit business manager for International TechneGroup Inc. (ITI; www.QFDcapture.com; Milford, OH). Add to this, says Larry Keeley, president of the design firm Doblin Inc. (www.doblin.com; Chicago IL), that many of the formalized NPD methodologies, like QFD, “reveal a certain kind of command-and-control bias.” He thinks that bias is part of the reason why U.S. automotive companies seem to be struggling “to get powerfully relevant products and services,” and suggests these tools “are specifically about making sure we focus on the things that matter for manufacturing. They do nothing to help us think through little curiosities, like the customer and what he or she might want.”

But since the ‘90s, QFD seems to have “evolved to its most appropriate applications,” comments Bob Hayes, vice president of product development for New Product Innovations, Inc. (NPI; www.npi.com; Powell, OH). In fact, QFD has changed. “Enough consultants [are] pushing the idea that you’re trying to make better engineering decisions,” says Finn. Another change is in the QFD methodology and its associated software.

 

Start with customers

QFD is a quality method consisting of several management and planning tools, explains Glenn Mazur, executive director of The QFD Institute (www.qfdi.org and www.mazur.net; Ann Arbor, MI). These tools are “a series of tables and charts to then systematically disaggregate customer needs into prioritized product requirements, functions, technology, systems, subsystems, parts, reliability, cost, manufacturing, production, equipment setup, operator training, and process controls.” This series constitutes a recursive decomposition process. It starts with a matrix of customers needs that gets decomposed and cascaded into more focused and detailed matrices, ending with a matrix that lists actions—the design/engineering specification.

Along the way, the NPD team attaches relative priorities to the statements of customer needs. That is part of the “voice of the customer” investigation. For example, says Kenneth Crow, president of the QFD consulting firm DRM Associates (www.npd-solutions.com; Palos Verdes, CA), having a cupholder is a nice-to-have feature; having good gas mileage is a more important feature. “While these are related to totally different systems and different levels of importance within a vehicle, you need to have that distinction so you can plan what your requirements are.”

One of the fundamental matrices in QFD is the House of Quality (HoQ). This matrix relates customer requirements to business objectives (and engineering actions). However, says Finn, “drawing the HoQ is not the point. The point is in analyzing the content and the data. A lot of people draw the HoQ just to get their manager off their back, and then they do what they want.”

However, don’t brush off that customer voice part. “QFD is a detailed process that can be arduous to put together for a product,” says Shawn Dodson, NPI’s director of quality and information technology. “You really can’t start QFD until you have the true voice of the customer.” There are lots of ways to do that. Keeley ticks of a couple of the “hundreds of structured analysis techniques”: agent-based modeling (dynamic optimization from past data), ethnographic research (studying customers), experience models (“living a life”), and focus groups. Add to this the Customer Needs Map, a Six Sigma tool that, explains Dodson, is boilerplate of sorts that makes the NPD team consider all the things a customer might want in a product. Ziba Design, Inc. (www.ziba.com; Portland, OR), also researches the backgrounds of users to understand what they grew up with (such as the movies, television shows, and even the commercials they watched), and to tap into the “reminiscent qualities” that people hold dear. Marty Gage, principal for design firm Rocket Surgery (www.rocketsurgery.biz; Columbus, OH), includes in his customer research looking for patterns in participatory exercises.

These tools are designed to get people to articulate what they wish for, and, says Eric Chan, founder and president of ECCO Design Inc. (www.eccoid.com; New York, NY), to “connect the emotional to the consumer.” Take the Toyota Prius, continues Chan. “The Prius is meaningful to its owners. It’s not [just about] saving money or saving gas. They feel they’re responsible; they want to be better, intelligent individuals.” So the question to figure out, Chan continues, “it’s not just [about] a beautiful product, but why is it beautiful?”

Two other formalized NPD methodologies should be mentioned. Design for Six Sigma is another structured learning process relating customer needs to quality to manufacturing, and so on. Another tool is failure modes and effects analysis. FMEA helps designers identify potential problem areas in a product and the corrective action to reduce the effect of that problem.

Regardless of NPD methodology, software is not required. Steve McCallion, Ziba’s creative director, gives this analogy: Hollywood has computer programs that can write a screenplay, “But it won’t be magical. It won’t be amazing.” Strictly logic-based approaches are missing that magical aspect, what McCallion calls “informed intuition.”

Therein lies the rub. Informed intuition is basically a manual evaluative process. Hayes echoes this point. “You can reduce [NPD] to the totally formulaic. However, there is this human element that has to come in, some interpretation of what was observed, heard, and seen built by customers.”

That all said, capturing, prioritizing, and analyzing customer needs typically involves lots of cutting and pasting and typing, things that software performs quite speedily.

 

New, improved QFD

Over the years, QFD has been improved. For starters, more emphasis has been put on the voice of the customer—”the qualitative front-end analysis,” says Mazur. Next, some of the time- and resource intensive parts of QFD have been made optional or simplified. For example, Blitz QFD was specifically developed by the QFD Institute as a “rapid approach to applying QFD to only the most critical customer needs,” explains Mazur. The approach “uses several newer and smaller tools upfront to better understand the voice of the customer and to uncover unspoken customer needs.” HoQ is not required. What’s more, he says, “We fixed the math.” Traditional QFD performed “improper” calculations using ordinal scale numbers. (For example, coming in “first” or “second” does not necessarily mean that the first-place winner is twice as good as the runner-up.) QFD is now based on ratio-scale numbers, which enables proper and relevant analysis.

Last, QFD users and consultants are moving away from the cookbook approach to applying QFD. Now, QFD is being customized for the user organization; consultants are extracting the QFD tools and sequencing them appropriate for that organization and its products.

 

NPD software

Nowadays, NPD software does a lot more than cut-and-paste. For example, ITI’s QFD/Capture has tools to build a set of cascading matrices, starting with a source matrix (customer requirements) and finishing with submatrices (manufacturing requirements by product subsystem). If engineering changes are made at the submatrix level, the software ensures that none of the customer requirements listed in the source matrix are violated. Also, the software automatically translates priorities from one matrix to the others.

Embedded traceability functions let users trace backwards through the QFD matrices to see what customer requirements are violated if a requirement can’t be met. This becomes helpful if manufacturing or an outside vendor can’t meet the final engineering specification. If the customer requirement turns out to be relatively unimportant, then maybe a deviation from the engineering specification would be acceptable. When criteria or priorities change, linked data in upstream and downstream matrices are automatically updated.

In the past, users would manually inspect all the intersections in the QFD matrices for exes, circles, triangles, diamonds, and such—symbols that link the relevance of customer requirements to engineering action. Now, the software can fish through the matrices for relevant engineering actions and add up the associated costs to determine a budget. Other tools determine the overall importance of customer needs, the coverage of requirements, and the market opportunities.

Another module in QFD/Capture is basically a web-enabled directory system for communicating between users in the supply chain. This module serves two functions. One, users can share QFD charts through the web. Second, the module can archive the results of customer surveys, focus groups, and so on, and then populate the source QFD matrix. “It’s a way to automate some of the tedium in the QFD process,” adds Finn. There are other QFD software packages on the market. However, admits Finn, “our biggest competitor—95% of the market—uses Microsoft Excel.”

Ultimately, how does one translate the mass of data regarding customer needs into design/engineering requirements? At times, admits McCallion, “it’s a very messy process. It’s a combination of art and science. We push the science part as much as far as we can, recognizing there’s an aspect of art as well.”