During my seven years as a visual designer at Cooper, I’ve learned that designing for complex digital products and services requires input from a number of unique perspectives to be truly effective. Furthermore, each of those perspectives must be effectively integrated throughout the lifecycle of the design process to achieve results that are consistently and predictably usable, useful and desirable.
In the course of managing, consulting and teaching, I have not only had the opportunity to define and discuss design process with my colleagues here at Cooper, but with countless practicing designers from organizations all over the world as well. Unfortunately, my observation has been that even when all of the right people are involved, more often than not, the various design disciplines opt to compartmentalize the problem. In other words, they divide the project into an interaction design problem, a visual design problem, and an industrial design problem. Each of these problems is then tackled separately, and the resulting individual design solutions get bolted together at the end. It’s a Tower of Babel situation, where huge opportunities are lost because the team fails to work together to come up with an innovative product solution and to employ a single, unified design language.
A fractured process makes for a fractured user experience
In practice, people view their experience with a product in a unified way. For example, when a user interacts with a cell phone, she doesn’t experience the phone’s behavior separately from the visual and tactile presentation of that behavior on screen and through the physical form factor. Why, then, don’t product teams consider these aspects of the experience in a unified way when designing solutions?
Of course, we know that many digital products and services represent extraordinarily complex, large-scale design challenges. A significant degree of rigor and a rational approach to methodology is required to bring together the diverse perspectives of the different design disciplines in a way that results in optimal creative abrasion, rather than destructive friction that threatens to bring the entire process to a grinding halt(1).To this end, let me share with you a few of the insights that we’ve gleaned from practicing a truly convergent approach to design.
(I should note that in an attempt to keep the length under control, I’ve focused this article on the convergence of interaction and visual design for products with defined hardware, like PC’s or handsets. We’re looking forward to sharing our experiences with integrating these two disciplines with industrial design in a future article.)
All for one, and one for all
First, each design discipline must be involved throughout the entire lifecycle of the design process in order to get full value from a convergent approach. This doesn’t necessarily mean full-time participation by all team members at all times, but it does mean consistent and regular involvement from the beginning.
A common approach that I have observed in the industry is to have research and analysis specialists conduct an investigation and deliver a document to interaction designers, who then develop and deliver wireframes to visual designers, who subsequently skin the wireframe for delivery to developers. Meanwhile, the industrial design team will have taken that same research report (or, in some cases, an entirely separate research report!) and developed a hardware solution to the physical components of the problem for delivery to manufacturing. At this point, the manufacturer is responsible for merging the software and hardware into a single, “integrated” product. This fireman’s brigade approach to design ensures that many good contributions get dropped between handoff points, never making it into the final design deliverable.
Industrial design shouldn’t happen prior to interaction design, and visual design shouldn’t occur after interaction design is complete. Rather, activities from each discipline should happen in parallel, with all design team members working in partnership on key activities. This includes research, requirements definition, the definition of the design framework, and the refinement and documentation of detail design.
The practice of interaction design tends to be highly process driven, while visual design is often less so. Yet, I have observed that a well-defined process improves the results of both. For example, a process that provides visibility, predictability and supports rational decision making minimizes the effects of subjectivity and the need for experimentation in order to achieve outstanding visual design solutions. Besides, a well-considered process is essential to the successful integration of the disciplines.
Find a process that brings the right constituents together
You may be familiar with Goal-Directed Design. The basic idea is simple: by focusing on accomplishing goals, instead of specific product features or technologies, we create the opportunity for discovering breakthrough design ideas. While the level of effort and certain tactical details may vary, for nearly every design challenge the same basic strategic approach proves to be most effective.
For instance, understanding and defining the design problem and the people it relates to before you attempt to develop serious solutions is always a good idea. Investing time in considering the fundamental form and key relationships of a design concept before attempting to resolve precise details is essential for avoiding painful, expensive and sometimes unworkable course corrections. Design considerations don’t suddenly evaporate when implementation begins. These truths are relevant to interaction design, visual design, and industrial design. This is why they are represented in the Goal-Directed Design process: Research, Modeling, Requirements Definition, Framework Definition, Design Refinement, and Development Support.
Convergence during design research
During research, the team’s focus is on interviews with stakeholders, subject matter experts, customers, and end users. This ethnographic-style research facilitates vision alignment with project stakeholders and provides critical data for the design process. At Cooper, we believe that it is essential for designers to be directly involved in the research. While research specialists exist, my observation has been that it is difficult for interaction designers to do their jobs well if they are not directly involved in the collection of the vital information that feeds the design process.
With this in mind, it is not uncommon for visual designers and industrial designers to be left out of this activity altogether. In some cases this is done to reduce cost, since even the interaction designers are fighting to justify every single interview. In other cases, the value of their participation is not well understood. It is true that some of the information from research activities can be shared in the form of a written report or through verbal communication. However, excluding key design team members from the activity entirely drastically reduces their ability to contribute to the design process, especially so for a convergent approach. Without direct participation, visual designers and industrial designers are forced to either accept the interpretation of those that conducted the research (often the interaction designers) or reject that interpretation entirely and rely on a purely self-referential design approach.
Another factor to consider is that each design discipline has unique information needs for effective design decision making. When one of these design team members is excluded, it is almost certain that some of this unique information will not be properly emphasized (or collected at all, for that matter). For stakeholder interviews, there is often a need to speak with exclusively visual or industrial design-oriented stakeholders, who may be disregarded with a purely interaction design oriented research approach.
Involve visual and industrial designers in stakeholder interviews
Key visual design stakeholders, for example, include GUI developers responsible for implementing the design, product managers or others who are responsible for the overall vision of the product and have information about the intended display, and senior corporate marketing people or others responsible for brand strategy. Also, make sure to identify brand stakeholders who can speak beyond what is written down in the available brand documentation. A visual designer needs to know what has been done before and why, as well as what the future portends, not just what the current state is. This will facilitate the process and execution of a visual design language strategy that works.
During stakeholder interviews, interaction designers look for a product vision, information about the users and relevant markets, presumed project schedule, time, cost, and technology constraints, as well as any general concerns that might affect the success of the project. In addition to this, visual designers also ask about brand attributes the design should convey, existing guidelines and examples of branded artifacts, competition, and technical constraints that have a bearing on display such as target platform, resolution and color depth. A final point that should not be overlooked is that stakeholder interviews represent an opportunity to initiate relationships with key stakeholders. For these reasons, we make full-time participation in stakeholder interviews a best practice for our projects.
Involve visual and industrial designers in user interviews
It is also important to consider full design-team involvement in first-hand user research. However, this is an activity where it is usually difficult to justify full-time engagement by all, because there are some very real trade-offs beyond the standard cost considerations. For instance, cramming one or two more people into that tiny office cubicle can certainly present difficulties. In addition, interview time is always a limitation, and there typically isn’t enough time for all team members to get equal time (besides the fact that it can be intimidating to have four or five people barraging the interviewee with questions). We address this last issue by making a practice of identifying a single team member to drive the main line of inquiry, while the others restrict themselves to providing support and taking advantages of gaps in the conversation to slip in a question. I have also noticed that there is often a few spare minutes left at the end of each interview session where discipline-centric topics can be explored.
A convergent approach relies heavily on the productive tension created by the unique perspectives of the different design disciplines. If one or more team member is excluded entirely from any significant part of the engagement—but particularly the research phase—the balance of power is lost.
Consider this scenario: The design team is reviewing an early framework sketch for a consumer-oriented photo and video sharing interface. Based on past experience, the visual designer on the team has some concerns over the information density represented by the current design and the impression it leaves. The interaction designers respond by referencing the research, suggesting that the degree of density shouldn’t represent an issue based on the people they interviewed and observed. If the visual designer has no direct experience with that research, where does the conversation go from here? The visual designer is not an equal participant in the conversation. This naturally results in design solutions that miss significant opportunities to fully address more emotional and inspirational drivers, which are key for creating memorable interactions that build long-term customer loyalty and brand equity.
Balancing resources during research
So, what is the right balance for participation in the research? For stakeholder interviews, full-time participation by all contributing team members is the way to go. The content of these interviews is critical to the success of all of the contributing designers, and there is no more efficient method for ensuring that the information is received and understood.
For user interviews, where the trade-offs are more significant, a more judicious approach is warranted. For projects that are defined by a familiar context of use, such as a typical office desktop PC configuration, and a user population with no unique characteristics, one or two days of full team participation may suffice. In these cases, the interaction designers are typically best suited to serve as a proxy for the full team for the remaining interviews. It is crucial that visual and industrial designers be involved in defining and adjusting the interview topic list as the research progresses and that regular check-ins are scheduled to allow for these other team members to participate in the iterative synthesis of the research data.
For initiatives dealing with a more foreign context of use, like a surgical operating room, or users with unique considerations, such as diabetics who tend to have vision or motor disabilities like shaky hands, it can be easy to justify four or five days or more of full-team user research participation, depending on the scale and strategic significance of the project.
What challenges and opportunities might a visual designer or industrial designer identify from an interview context like this one?
How might a visual design language strategy differ for people in these different environments?
Convergence during Modeling
Based on the synthesis of the completed research, the design team models users (personas) and workflows in order to provide insight into user goals that will drive and focus design, and support effective communication with key decision-makers. Interaction designers typically begin this phase by understanding and modeling processes, and more importantly on driving the development of personas.
Visual designers and industrial designers should consult into the persona creation process to ensure that these essential user archetypes align with brand strategy, and more specifically, contain appropriate emotional and aspirational dimensions. When this perspective is omitted, the resulting persona set tends to over-emphasize the satisfaction of functional goals (what users want to accomplish) and leaves out experience goals (how they want to feel while they are using the product or service). Not having a strong perspective on experience goals can cause the team to overlook opportunities to deliver unique, memorable, and ultimately brand-building experiences. In addition, a minimal amount of participation in the process itself can ensure that the entire design team feels a sense of ownership, and has the chance to truly understand and internalize the key factors underlying the persona set.
Beyond the creation of user models, one of the crucial activities of this stage is achieving a shared understanding with stakeholders of the process, workflows, and other important concepts that will shape the decision-making process throughout the lifecycle of the project. In this regard a picture really is worth a thousand words. The right visualization can spark immediate understanding and healthy discussion in a way that other forms of communication can’t match. It is also true that the wrong visualization can confuse and mislead stakeholders. Visualization is powerful, and it can take considerable effort and time to realign understanding once stakeholders have an image in their heads. Since visual designers are often more adept at effectively visualizing complex processes and concepts, it is only common sense to involve them in this activity to ensure that communication is clear and effective.
Convergence during requirements definition
Ultimately, the objective of the research and modeling activities is to define and communicate a set of sound design requirements that will lay the groundwork for the design effort. To achieve this goal, interaction designers construct and explore idealized workflows (in the form of context scenarios for the personas), which illuminates a list of critical needs. This approach allows the design team to deal with big issues early in the process, before concrete design begins and it becomes too expensive and painful to address those concerns. While it continues to makes sense for visual and industrial designers to consult into this process, this is also the point at which it is important to initiate a parallel effort to define appropriate visual-design and industrial-design oriented requirements. Some of these requirements are typically more technical in nature, such as type size, screen resolution, and bit-depth (for visual design), display size, device weight, and battery life (for industrial design). Others focus on the more emotional and aspirational brand qualities the design language should convey.
A brand is the embodiment of all the qualities people associate with your company, product and or service. At the end of the day, it is what people remember about their experiences that matters. While many people limit the concept of brand to a logo, it is important to remember that a logo is only one representation of the attributes that define a brand. It is like the desktop shortcut on a PC that serves as an entry point to the larger experience.
What attributes do you associate with these brands, and what effect does that have on your behavior?
Brand attributes are an important factor in establishing a complete set of design requirements. Many organizations make considerable investments in defining, establishing and maintaining these attributes, given that they are a key to building brand equity. By not incorporating brand considerations into your work, you are failing to tap that equity, which can be a powerful force in setting initial expectations and impressions, and moreover, you lose the chance to add to that equity. With that said, while they should have significant influence and effect on any design language strategy, corporate brand attributes are typically not specific enough to effectively drive detailed design decisions for an individual digital product or service. Corporate brand attributes are designed to drive communication across the entire brand platform, which is made up of a plethora of brand applications of all shapes, sizes and materials. Digital products in particular are almost always insufficiently considered in brand guidelines.
Furthermore, the users of a product or service can be distinct from the audience that the corporate brand is designed to serve. In order to ensure a strategy that will guide and drive good design decision making, we develop an explicit set of user experience attributes that are a synthesis of how the organization wants to be perceived (the corporate brand), how the users want to feel about their experience (personas’ experience goals), and other factors specific to the product or service to be designed, such as the influence of competition and opportunities to differentiate.
Visual design: defining experience attributes
The process for establishing a set of experience attributes starts with the designers synthesizing the information from stakeholder interviews and existing brand documentation into a list of terms that represent patterns in how the ideal product or service is perceived. We’ll often fill a whiteboard full of these recurring terms, at which point a process of consolidation begins. Clear patterns begin to emerge around a few key concepts, like Trustworthy or Innovative, which often represent an intersection of the different perspectives.
Key patterns are identified in the way the ideal experience is described, which forms the basis of concept categories.
These key concepts become labels for attribute categories, under which related terms can be organized. It’s rough at first, and it sometimes takes a few attempts to get a collection of categories that resonate, but it is at this point that the elemental experience strategy begins to form. From here, the categories are refined into an ideal set of attributes and supporting terms, based on their fitness for driving design language decisions.
Not just any term will do when it comes to useful experience attributes. The body of research, and specifically the work that Byron Reeves and Clifford Nass published in “The Media Equation,” suggests that the way we approach and appreciate our interactions with computer-based products and services is surprisingly similar to the way we interact with other human beings. For this reason, some of the best attributes for shaping user experience are words you might naturally use to characterize a person, such as smart, funny or humble.
Generally, it is helpful to think of attributes in terms of functional (those we can see and that define how well the product or service works) and emotional (those that are more abstract, and are often largely internal or invisible and most responsible for creating an “image”). High quality is an example of a functional brand attribute, while trustworthy might be considered an emotional attribute. It takes an emphasis on both to establish the highest value brand. In my experience, too much emphasis on functional attributes within the design language strategy can lead to a brand that is perceived as a commodity. I find that functional attributes are helpful in establishing the basis for making a brand promise, but they rarely represent the unique differentiating factor that drives long-term customer loyalty on their own. In this way, functional attributes could be considered hygienic. People expect these attributes, look for them when establishing their first impressions, and certainly miss them if absent from the language, but functional attributes are almost never the thing that makes a product exceptional and memorable. This is where emotional attributes come in. They go beyond the mechanical satisfaction of functional goals and connect with people on a deeper level. It’s the difference between a high-quality cup of coffee and a Starbucks.
At the end of the day, as designers, we want to look for a set of attributes that tells a comprehensive story that resonates with both stakeholders and users, and has a healthy amount of tension which will be productive for exploring and establishing boundaries. For example, the attributes innovative and mature create some natural, productive tension. This contrast establishes a continuum that we can explore between the attributes, while establishing an extreme for each opposing attribute. For instance, a design language that takes the concept of innovation to the bleeding edge can no longer be considered stable and mature, and therefore falls outside the boundaries of the strategy defined by the overall attribute set.
A set of four experience attributes, along with their supporting terms. A good attribute set always contains productive tension that is good for establishing the boundaries of a design language strategy.
Along these lines, it is often as productive to describe the negative space as it is to discuss the positive. In other words, make sure you spend some time discussing and explaining what it is categorically not, as well as what it is. For instance, the product should be brilliant, but not “bleeding edge.” An ideal negative attribute is one that represents a good thing taken to its extreme, rather than an inherently negative concept. For all experience attributes, but especially for expressing the significance of negative attributes, providing a visual reference can be very effective.
The completed attribute set forms the basis for a visual (and tactile, if industrial design is involved) language strategy, which is vital to defining essential design requirements. While the experience-attribute set is certainly influenced by, and has an effect on, the behavioral aspects of the design language that the interaction designers are driving, it generally serves as a tool for guiding visual design decisions. It defines boundaries for exploration and provides a set of metrics to judge the fitness of design directions.
Still, it is important that interaction designers participate in the same way visual designers should contribute to the development of personas. Interaction designers represent a good resource for resolving any apparent contradictions in the research findings. Furthermore, because interaction designers often have the most direct exposure to the research, they are well equipped to contribute to the refinement of individual attributes as well as to test the general fitness of the attribute set as a whole. Perhaps most importantly, involving the interaction design perspective in this activity ensures that any disconnects in strategy between the two design disciplines gets resolved early and the overall vision remains aligned.
Convergence during framework definition
Once agreement has been reached on the fundamental product definition, the user targets, and the appropriate design requirements (including the design language strategy), the next step is to frame a solution. This involves creating high-level sketches and studies that allow us to validate the major concepts and general design direction quickly and inexpensively. Designers translate the goals defined by the user and domain analysis into a concrete vision that can be inspected, discussed and evaluated. The ultimate objective of this phase is to establish a clear, uncluttered structure that will provide a solid and stable foundation for designing and refining the details needed to implement the design.
One of the single greatest reasons for design projects failing, in my observation, is a failure to first establish a design framework before attempting to resolve the details. It’s not all that different from trying to design a skyscraper without placing proper emphasis on the foundation and superstructure. You can’t start over once the skyscraper is half built, and it can be just as painful to revisit a fatal flaw in the design framework once a significant amount of detail design has been done. The excuse that is often given is that there isn’t enough time; “Our developers are waiting, we need to move quickly!” Of course, there is no advantage to going one-hundred miles an hour if you are going in the wrong direction.
Exploring the visual language system
During the framework phase it continues to make sense for the different design disciplines to work in parallel, with regular opportunities to get input from the entire team. While interaction designers are focused on establishing a behavioral framework, relying on scenarios and personas to rough out the fundamental interface panes, views and navigation that will be required, visual designers must similarly begin exploring a visual language framework that will address the strategy, underpinned by the experience attribute set. This starts with the development of visual language studies, which combine the basic visual language building blocks of color, type and style in a thumbnail-like rendering. The intent of these expedient studies is to present specific visual language directions that map to the overall strategy. The studies are a catalyst for early discussion and decision making. The various directions are vetted with the entire design team, and then shared with key stakeholders as part of a facilitated discussion aimed at narrowing the focus for further exploration and refinement.
An example visual language study set
One of the keys to this approach is maintaining the discipline to separate the emergent behavioral framework from the visual language discussion, which may seem counterintuitive; after all, aren’t we advocating a convergent approach? Let me explain. In most cases at this point, the interaction design is not mature enough to provide a stable platform for applying visual design. In other words, the simple “rectangles” that interaction designers use to expediently explore the range of relationships within the system as a whole do not offer a level of granularity that is sufficient for applying visual design. It is nearly impossible to apply visual design to approximate form in a meaningful way. To force more specificity at this stage can be perilous. While the team might have a hypothesis for certain behavioral constructs, it is almost guaranteed to be wrong in some number of ways.
An interesting thing happens when you apply visual design to interaction design; it becomes real. It is like wrapping a skyscraper’s superstructure with glass and the other exterior materials. The design is no longer viewed as a framework, where evaluation is focused on concept, approximate form and general direction. The details now demand recognition. As with concept and workflow visualizations earlier in the process, once a specific image is formed in stakeholder’s minds, it can be very difficult to change that initial frame of reference. You do not want visual designers to establish a separate, competing interaction design framework.
So why not just wait for the interaction design framework to mature? The simple answer is that doing so will put the visual design effort out of sync with the interaction design effort, limiting and even eliminating opportunities to benefit from convergence going forward. Once the team gets out of sync, it is very difficult to get back into alignment without significant cost. Visual language studies represent color, type and style treatments applied to a generic interface structure. The objective is to produce a sort of tailor’s dummy, which is detailed enough to apply visual design to, but that doesn’t represent any specific interface recommendation. The entire team should collaborate on the approximate form and the primary elements that make sense to represent within this structure. If the design team knows that the design is likely to be oriented around data grids and tables, then it makes good sense to include that as an element for treatment in these initial studies. The key is to ensure that the studies drive a discussion of visual language, and do not encourage discussion of behavior. An effective way to further promote this, in addition to crafting the right underlying framework, is to crop the studies for presentation. This further abstracts the behavior, and helps to prevent stakeholders from confusing recommendations for visual language with interaction design.
On average, it takes somewhere between three and five unique directions to explore and discuss the range necessary to adequately address the established visual design requirements. Bear in mind, to produce even just three compelling, viable studies, it is common to explore as many as ten or more rough concepts. For projects where the visual language and experience is crucial, such as with many consumer applications, a wider exploration may be justified, where a straightforward, internal business utility will likely require less. Another helpful tip for establishing an effective set of studies is to include one or two directions that push the envelope. Like negative attributes, defining the negative space tells us a lot about that which is within the boundaries. With that said, we never present studies to stakeholders that the design team doesn’t think are viable. Remember the designers’ corollary to Murphy’s Law: if there is a design direction you don’t want stakeholders to pick, that is the one they will pick.
I occasionally get the question, “why not just show a single recommendation?” There are a couple of reasons why this is not the best approach for this material. First, visual language in the context of complex digital systems is highly nuanced. There is almost always a range of reasonable and appropriate ways to express the visual design requirements, in contrast to interaction design, where scenarios tend to suggest a more narrow range of possibilities. Secondly, visual language studies are not meant to suggest final, detailed recommendations. Their intrinsic value is in the discussion that they inspire with stakeholders. They are intended to present strategy in a format that can be pointed at and questioned.
You might think of visual language studies as a visual creative brief. Remember, the objective for visual designers during the framework phase is to map out the boundaries of the design solution. The stakeholder discussion that visual studies feed permits the design team to sharpen their understanding of the visual design requirements, and therefore narrows the focus of exploration needed to achieve a convincing solution. In other words, it reduces the need for trial and error, which increases predictability and allows the visual design effort to share a schedule with the other disciplines, which is necessary to maintain a convergent approach.
Collaboration during visual language study creation
All design team members review and provide feedback on the visual language studies prior to presenting to stakeholders. When presenting the studies, we’re looking for gut reactions first. Ask stakeholders to consider their first five-second reaction. Does the study align with their vision for the product? Will it resonate with the targeted users, defined by the persona set? Does it integrate with the overarching brand strategy? Start with a quick review of the material you used to drive the studies, and around which you have previously established consensus.
Next, ask stakeholders for more considered reactions, with a focus on the experience attributes and strategic implication, rather than personal preference. It is helpful to explain how the use of color, type and style relate to the experience attributes, as well as the overall strategy that the study is emphasizing related to the overall attribute set.
Ultimately, we’re looking to build consensus around one or two directions to explore and refine further. While directions often primarily correlate to individual studies, they may comprise aspects of multiple studies. Remember, it’s less about selecting a specific study and more about gaining a shared understanding around which directions are worth investing additional effort in. An excellent way to make consensus easier to achieve is to work with the group to eliminate the outliers. By eliminating one or more studies, perhaps by agreeing on which directions are too progressive versus too conservative, you narrow the range of options and move closer to consensus.
Converging with interaction design
As the interaction designers progress through the framework stage, exploring a range of key path scenarios, a few archetypal screens begin to take shape, where there is sufficient detail and confidence to begin the process of integrating the visual and behavioral languages. By the time the team moves to detail design, we want to be dealing with a single unified design language, including industrial design if applicable. This process begins by applying the visual language strategy to one or two interaction framework sketches. The result is an initial design language archetype.
Visual language study + interaction framework = design language archetype
This is the earliest point in which the various perspectives of the different design disciplines can be effectively unified into a single vision. Within the design language archetype, the visual design must not only respond to the defined visual-oriented requirements, but perhaps even more importantly, it must effectively support behavioral imperatives as well. How are visual affordances and cues utilized to clarify behavior, define relationships, and establish clear information hierarchy? Because the effective presentation of behavior has such a fundamental impact on shaping positive long term experiences, our general philosophy here at Cooper is to emphasize it in our approach. By integrating distinct interaction and visual design perspectives, relatively early in the process, we are able to ensure that the right behaviors are expressed in the right way, so that they are not only clear and understandable but unique, memorable and connected to a larger brand experience.
Onward to design refinement
I’ve found that it’s the activities described above that either happen out of sync or not at all in most design projects. This lack of convergence can doom a project from the beginning, making it painful (or impossible) to pull together a seamlessly integrated user experience later on.
During design refinement, the visual designers, interaction designers, and industrial designers spend more “alone time” working through the details of the now common vision. It’s easy for individuals to get too heads-down working on their own area, so be diligent about creating opportunities for the team to collaborate and share progress with each other outside of formal collaboration time and internal reviews.
As a final thought, truly integrating visual design, industrial design, and interaction requires more than just skilled, compatible designers. It also requires a commitment from product and project managers, who must allow time in the schedule for collaboration while also maintaining the separation that allows individual designers the space they need to bring their expertise and perspective to the table.
1. The Creative Priority: Driving Innovative Business in the Real World by Jerry Hirshberg (Hardcover – Jan 20, 1998)