design
inquiry
Robust Design Review Conversations
Andy Dong
Massimo Garbuio
Dan Lovallo
Background
The decision to take a product from its conceptual design into detailed design has properties of strategic decisions as defined in the strategic management field. The decision irreversibly commits a significant investment of resources (high degree of commitment) toward delivering or expanding a new product or service (changes the scope of the firm) (Shivakumar, 2014). Researchers and scholars in engineering design and new product development have, as such, motivated their research in decision-based design as having relevance to the strategic nature of these decisions. To improve the quality of design decisions, scholars of design decision making have tended to focus on how to take a decision for a range of tasks consistently faced in new product design and development (Krishnan & Ulrich, 2001). Perhaps the most important decision taken during design is concept selection, the analysis and evaluation of alternative concepts, leading to the selection or consolidation of one or more concepts for further development. A range of normative decision-making tools and methods for concept selection exist, including concept screening (Ulrich & Eppinger, 2004), pair-wise comparison charts (Dym, Wood, & Scott, 2002), concept scoring matrices (Frey et al., 2009; Pugh, 1981), multi-attribute utility analysis (Scott & Antonsson, 1998; Thurston, 1991), and Pareto dominance (Malak & Paredis, 2010). As a consequence, there has been a very robust and long-standing debate surrounding the decision method to use once a discrete set of alternatives is known (Frey et al., 2009; Hazelrigg, 2010; Reich, 2010) and the type of design decisions for which the axioms of decision theory ought to be applied (Hazelrigg, 1998; Thurston, 2001).
Lost in this debate, though, is the quality of the decision-making process itself. Taken together, a broad body of research in the strategic management literature points to the conclusion that decision processes matter to the performance of the project first and to the performance of the firm second (Fredrickson & Mitchell, 1984; Papadakis & Barwise, 2002). More recently, two studies have pointed toward the importance of conversations over numbers and financial analysis in decision making. First, in a study of new-to-the-firm products, resistance was won by using micropolitical strategies. This happened by building a coalition of supporters but especially by framing the product in terms of the firmâs existing products, strategies, and competitive thrusts (Sethi, Iqbal, & Sethi, 2012).
Second, a large sample study of strategic decisions has highlighted how strategic conversations are substantially more important than the financial analysis of a decision in shaping the outcomes of such decisions (Garbuio, Lovallo, & Sibony, forthcoming). In this study, it was âhowâ the executives talked about the decision and its underlying assumptions that had an impact on whether expectations in terms of market share or profitability were met, not âwhatâ financial analysis was performed.
Building on research on the quality of design dialogue in accomplishing actions and practices (Luck, 2009) that enable the emergence of tangible goods (Dong, 2007; Oak, 2011), this study aims to contribute to the decision-making scholarship in design by investigating the quality of design review conversations. We focus on the situation of the review of design concepts presented throughout a junior-level (third-year) undergraduate industrial design course and the final presentations of an entrepreneurship course at a public university in the United States. The conversations in the industrial design course contain discussions about multiple design concepts, which can lead to the abandonment or further development of design concepts until a final concept is chosen. In contrast, the entrepreneurship presentations communicate a single project and are representative of the type of presentation to an executive committee tasked with making a resource allocation decision (i.e., a go/no-go decision).
Theoretical Frameworks
The evaluation of a design concept is a key part of the design process. By evaluation, we mean assessing the merits and shortcomings of proposed design concepts (e.g., non-fully elaborated ideas for new products), which takes place throughout the design process until a single, fully elaborated candidate design is selected as the final option. To assist designers in filtering concepts, researchers have proposed creativity metrics (Nelson, Wilson, Rosen, & Yen, 2009; Oman, Tumer, Wood, & Seepersad, 2013; Shah, Smith, & Vargas-Hernandez, 2003; Verhaegen, Vandevenne, Peeters, & Duflou, 2013). The problem we see is that these evaluation metrics call for deductive reasoning, such as in quantifying novelty by comparing an idea to a universe of ideas (Maher, 2010; Shah et al., 2003). If the idea is the designerâs own, then the designer may prefer it to others even when there is no rational basis for the preference (Nikander, Liikkanen, & Laakso, 2014). Empirical research in industry for concept evaluations also describe decision makers as tending to apply variables amenable to deductive analysis including product timing, staffing, and platform when evaluating innovative projects (Krishnan & Ulrich, 2001; van Riel, Semeijn, Hammedi, & Henseler, 2011). Even in the situations when the concept is in its early phases, evaluation techniques employ highly deductive analysis requiring a substantial amount of criteria for analysis (Ulrich & Eppinger, 2004).
If the purpose of design evaluations were to evaluate concepts only as presented with no further elaboration possible, then these types of metrics make sense. The accepted practice is that design evaluation per se in the selection phase (i.e., when decision makers are presented with a discrete set of options) should only examine the merits of options or âmerit-based evaluation.â However, we believe that design evaluations should always entail both the evaluation of the quality of the design concept and be âforward lookingâ for âwhat might beâ or âopportunity-based evaluation.â
Thus, rather than the evaluation of a design concept being âstatic,â based only on existing evidence, we propose a dynamic model. We hypothesize that a robust design review conversation should consist of at least two components. The first is strategic analysis. We define strategic analysis in the design context as the extent to which decision makers use evidence to evaluate design quality based upon a priori design criteria such as the requirements. When we refer to evidence, we mean propositions that justify a belief; propositions may include inter alia:
â˘Observable properties of the concept, such as physical characteristics.
â˘Arguments based upon belief or experience, such as professional standards.
â˘Secondary data, such as consumer preference data.
â˘Claims, such as conclusions drawn from prior evaluations of the design concept.
We hypothesize that the second component of a robust design review conversation is the quality of generative sensing. We define generative sensing as the process of creating new hypotheses to explain, resolve, or challenge the evidence in favor of or against a design concept, evidence that was itself generated from an evaluation of the design concept. Generative sensing based upon the output of the evaluation of the design concept can lead to new knowledge that changes the designerâs view of the design concept, resulting in a reframing of the problem itself (Dorst & Cross, 2001). In design, making the leap from the evaluation of a design concept to a final design concept is not solely about testing the merits of the design concept as a fait accompli. It is about generating a series of tests of the design concept until an appropriate concept is identified. In the context of design evaluation, generative sensing entails inferences to explain the evaluation. These inferences may provide resolutions to problems identified by the evaluation when the evaluation is adverse. In contrast, a positive evaluation may spur the proposition of conditions that would undermine the basis of the evaluation in order to test the robustness of the evaluation.
Our concept of generative sensing shares some ideas with the concept of the primary generator (Darke, 1979). A primary generator is a conjecture, or better stated, a scheme based upon a value judgment as the basis for generating potential solutions. The value judgment, which does not satisfy all constraints, provides a âway in to the problemâ (Darke, 1979, p. 38). Generative sensing entails producing hypotheses that may resolve (or further expand) issues encountered in the evaluation of a design concept. Thus, rather than a âway in to the problem,â generative sensing can be seen as creating alternative âways through the problem.â
The form of logical reasoning underlying generative sensing is abductive reasoning. The concept of abduction in design is philosophically very powerful as it introduces a mechanism of discovery through a form of logical reasoning. Scholars have theorized that the relevant form of abductive reasoning in design is innovative abduction. Innovative abduction produces an explanation (the design concept) for the desired value, the function, and, in turn an explanation (the form) for the design concept (Kroll & Koskela, 2014; Roozenburg, 1993). As Dorst writes, designers must engage in a form of reasoning âto figure out âwhatâ to create, while there is no known or chosen âworking principleâ that we can trust to lead to the aspired valueâ (Dorst, 2011, p. 524). The term âvalueâ is not restricted to economic or financial value, but, rather, any values to which the designer aspires (Friedman & Kahn, Jr., 2003; Le Dantec & Do, 2009; Lloyd, 2009). In other words, abductive reasoning in design generates hypotheses that, if true, would explain the form of the proposed product and its mode of operation to achieve a desired value (Roozenburg, 1993). Design theory scholars propose that the major premise that abductive reasoning must infer is the rule that connects a form to its function within an operating environment (Zeng & Cheng, 1991). This logical reasoning from function to form appears to refer to Sullivanâs widely cited credo that âform ever follows functionâ (Sullivan, 1896), although scholars of abductive reasoning in design do not refer to Sullivan explicitly. If function or value is intentional, then innovative abduction in design is about inferring a form that achieves an intended purpose. The purpose may not necessarily be utilitarian or performative.
Roozenburg (1993) introduces the following notation to describe innovative abduction:
q | a given fact (function or value): q |
p â q | a rule to be inferred first: IF p THEN q |
p | the conclusion: p |
Kroll and Koskela (2014) extend the model of abduction proposed by Roozenburg (1993) and Dorst (2011) into a two-step recursive inference of the innovative abduction: the first step involves abduction of a concept given a function and the second step involves abduction of a form given the concept inferred from the previous step.
q | a given fact: function |
p â q | first conclusion: IF concept THEN function |
p | second conclusion: concept |
q | a given fact: concept |
p â q | first conclusion: IF form THEN concept |
p | second conclusion: form |
We propose that the process of design does not (should not) arbitrarily stop. In other words, the participants should continue to propose hypotheses that infer the link between function and form in a recursive manner. Each inference is only a partial resolution of the design problem, the depth of which depends upon the complexity of the problem and the number of subproblems to be resolved (Zeng & Cheng, 1991). Thus, inferring the working principle (concept), which is comprised of mode of operation and way of use (Roozenburg, 1993), can entail multiple recursive inferences....