Toward Identifying The Inventive Templates of New Products: A Channeled Ideation Approach

Reprinted with permission from the Journal of Marketing Research, published by the
American Marketing Association, Goldenberg, J., Mazursky, D., & Solomon S., 36, 200-210.
JACOB GOLDENBERG, DAVID MAZURSKY, and SORIN SOLOMON*
New product ideation might be improved by identifying and applying certain well-defined
schemes derived from an historical analysis of product-based trends, termed “templates.”
These templates might contribute to the understanding and prediction of new product
emergence. The authors derive templates in a study that maps the evolution of product
changes by adapting a set of intrinsic operations originally designed to uncover hidden
logical patterns in technological inventions. They find that the majority of new product
versions can be accounted for by as few as five templates. The authors define the five
templates and show that they derive from six elementary (first principle) operators. A
procedure for utilizing the dominant template, termed “Attribute Dependency,” is outlined,
followed by a report of two experiments examining its usefulness in the context of product
ideation.
TOWARD IDENTIFYING THE

INVENTIVE TEMPLATES OF NEW
PRODUCTS: A CHANNELED
IDEATION APPROACH
*Jacob Goldenberg is a lecturer, School of Business Administration (e-
mail:msgolden@mscc.huji.ac.il), David Mazursky is a senior lecturer, School of Business
Administration (e-mail: msmazur@mscc.huji.ac.il), and Sorin solomon is a professor, Racah
Institute of Physics (e-mail:sorin@vms.huji.ac.il). Hebrew University, Jerusalem. This research
was supported by grants from the K-mart International Center of Marketing and Retailing and
the Davidson Center. The authors wish to thank Josh Eliashberg, Zvi Gilula, Gita V. Johar, Mike
Kamins, John Kearon, Amnon Levav, Eyal Maoz, and Bob Shoemaker, for helpful comments on
a previous draft, and especially Don Lehmann for suggestions in various stages of this research.
The authors also thank Russ Winer and the three anonymous JMR reviewers for their support
and direction. The article has benefited from presentations at the Columbia, Stanford, and
Wharton business schools.
SIT LTD. 16 Tozeret Ha'aretz St., Tel Aviv 67891, Israel. Tel: 972-3-6070 480 Fax: 972-3-6070 481. www.sitsite.com
The invention of new products has traditionally involved methods that encourage the
generation of large numbers of ideas. The notion that the rewards of generating a large number
of ideas outweigh the costs can be traced back to early studies in the field (e.g. Marschak,
Glennan and Summers 1967). In view of the fact that this process tends to be highly complex
and unformalized (e.g. De bono 1992; Klein 1990; Woodman, Sawyer, and Griffin 1993), those
involved in generating new ideas may seek ways to become more productive as they progress
from one ideation task to another. Some may succeed in identifying patterns of invention that are
common to different contexts (e.g. Boden 1992; Dasgupta 1994; Weisberg 1992) and apply them
within a certain product category, or even try to apply them to other product categories.
Individuals who adopt such a cognitive strategy may expect to gain an advantage over others who
treat every task as new and unrelated to past ideation. However, even if they prove productive,
the patterns are likely to be idiosyncratic and, quite often, not even verbally definable (Weisberg
1992). As such, they lack permanence and generalizability. The main thesis of this paper is that
certain patterns are identifiable, objectively verifiable, widely applied, and learnable, and that
these patterns, termed templates, can serve as a facilitative tool that channels the ideation process,
enabling the individual to be more productive and focused.
Let us illustrate these notions with an example. Analysis of Domino’s Pizza’s promotion
strategy offering a discount for any delivery exceeding a 30-minute time limit shows that the
company created a dependency between two previously unrelated variables, namely price and
delivery time. The dependency was created by introducing a step function between the two
variables: Within the first 30 minutes following the order, the price remains constant and
immediately thereafter, it is discounted. This specific dependency could be applied in other
contexts as well. Similar discounts could be offered, for example, in supermarket or laundry
delivery services when a predetermined delivery time limit is exceeded. Thus, we can say that a
replicable pattern has been identified.
Templates as Product-Based Trends

We can now extend the notion of replicable pattern to the notion of template. Templates
represent replicable patterns that are generalizable across variables and products. Let us exemplify
this extension by illustrating how one of the five templates identified in the present study,
namely, the Attribute Dependency Template, is derived. Note that in the Domino’s Pizza
example a replicable pattern is obtained by duplicating the dependency between the two specific
independent variables (price and time) through a step function. A general rule that new product
ideas can be generated by introducing a dependency between two previously independent variables through a
step function is far more general than the mere duplication of a pattern as suggested in the cases of
the supermarket or laundry delivery services. This template and its generalizability across
products can be illustrated with an example drawn from a recent study by Andrews and Smith
(1996): Hungry Jack syrup bottles are designed for microwave oven use. The bottle labels change
color on reaching a certain temperature, thereby informing consumers that the syrup is ready.
Although, at first glance, the analogy may appear remote, this example contains the key features
of the same template: The two independent variables, in this case, are temperature and label
color. A dependency is created by a step function between these two variables. Up to a critical
temperature, the label color is not activated, and on reaching it, the color changes. The
generalized form no longer involves the specific variables of the Pizza example, nor does it
necessarily involve the same product (or service). Yet, it is identifiable and general across
products and services and, thus, it is defined as a template.
Since the operation of templates involves manipulating product attributes rather than
market parameters, they can be used in considering new market needs in currently
underdeveloped product markets. For instance, the aforementioned template might be applied to
generate a new idea in pizza delivery services by creating a meaningful dependency between two
different attributes, such as price and temperature. Accordingly, a company might offer a price
discount if the temperature of the pizza falls below a predetermined level at the time of delivery.
The added value of the template approach is that it draws on the identification of similar
structures in former product changes, providing a different angle and sometimes more accessible
resource for ideation, compared to the information obtained from analysis of current market
needs. For example, the idea of creating a dependency between pizza temperature and price may
originate from market-based information (i.e. information derived from the current market
environment, e.g. Griffin and Hauser 1993, such as that implied by research indicating that
temperature at delivery is a crucial factor in the choice of pizza service). Alternatively, it may be
derived from product-based information (i.e. information that is inferred from inspecting the
internal dynamics of the product, such as the incidence of dependencies between variables
among new food-delivery services). Although the price-temperature idea may evolve both from
market-based information and/or product-based information, the template approach may be
more accessible since it specifies a structured framework for obtaining such an idea.
The added value of product-based information is especially apparent in more complex
ideation contexts such as in the following example. A new idea in the category of drinking glasses
might be a glass with a dual insulation capability: When the temperature of the content is high the
insulation is low; when the normal drinking temperature is reached, the insulation is high. The
advantage of such a glass would be that the content would rapidly reach the optimal drinking
temperature which would then be maintained for an extended period of time. Relying merely on
market-based information that the two factors - rapid cooling and heat preservation - are
important for consumers could, potentially, lead to such an idea. However, relying on product-
based information by adopting the Attribute Dependency template may be even more useful:
The information about these relevant parameters and the desired dependencies is contained in
changes in products overtime. Other examples of an Attribute Dependency in this category are a
drinking glass with colored lower section to camouflage residual juices or extracts, and
transparent at the top, the dependency between height and diameter as manifested in cone-
shaped cups allowing compact storage (in stacks), or a glass with base wider than rim to reduce
the danger of spilling when carried around. This information, and hence, the template, should be
especially valuable when market-based information is less accessible or invalid.
Clearly, many important product inventions have been derived from market-based
information, using a variety of marketing research methods. However, the value of these
methods in offering a rich and exhaustive set of tools for ideation has been criticized (e.g.
Petroski 1994). Durgee, O’Connor and Veryzer (1996) noted that many wants lie below the
surface and that current product users are not able to express wants and needs for non-existing
products. Referring to the common method of asking buyers to describe problems with current
products (Crawford 1991), they noted that big leaps to new product ideas may not be uncovered
when respondents describe problems in terms of current products.
One possible outcome of excessive reliance on market-based information is the
disproportionate effort that is currently devoted to “me-too” products, namely products which
mainly involve product line extensions, improvement of current products, and cost reduction
(Wind and Mahajan 1997). In contrast, product-based information has been shown to be
effective in various other decision making contexts (e.g. reverse engineering, Srinivasan, Lovejoy,
and Beach 1997; and “product archaeology”, Ulrich and Pearson 1993). It is our contention that
such information will be similarly effective in the context of ideation.
Deriving and Formulating the Templates
An approach for analyzing the dynamics of evolution of technological innovations was
developed by Altschuler (1985). We examined it for suitability in the context of product
evolution. Altschuler conducted a backward analysis of problem-solution relationships and
succeeded in identifying about 200 common phenomenological patterns, which he termed
“standards”. Although Altschuler’s approach was aimed at inventive problem solving of
technological systems rather than inventing new products, it is conceptually consistent with the
present template approach in the sense that it focuses on internal operations. In the present
study, only those operations that are relevant to the context of products were examined and
adapted.
The initial identification of templates in our study was obtained by mapping structural
changes in the soap category. A preliminary condition for inferring templates is knowledge of the
chronological development of products. Existing data on the evolution of soap-related products
were used in describing the original product in each stage, and the transition to the new product
in the following stage1. The commonalties identified in the transition between products were
classified in relation to Altschuler’s typology, and were considered candidate product templates.
This procedure was then replicated for three other product classes, namely, hygiene products,
bank accounts, and sneakers. In these mapping studies five major templates were revealed. In the
following section, we formulate the template taxonomy. Several basic definitions required for the
development of the template taxonomy are provided below.
Definitions
The characteristics of a product can be divided into components (objects such as the legs
or seat of a chair) and attributes (variables of the product such as the color or height). It should
1 The description of the development of the soap related category is based on a research project conducted by
Brand Genetics Inc. (Research Report No. 96-01) and is presented here by permission from the company.
be noted that at the stage of template construction we consider only attributes that consist of
factual information. Abstractions and inferences (e.g. aesthetics, Chattopadhyay and Alba 1988)
are considered only at a later stage.
In addition, a distinction is made between internal attributes and components which are
under the manufacturer’s control and external attributes and components which are not
controllable by the manufacturer but are present in the neighboring environment of the product
in its common usage. Thus, the legs and seat of a chair are internal components while color and
height are internal attributes; floor and table are external components and the height and weight
of the person sitting on the chair are external attributes. Since all these components and attributes
belong to the product and its neighboring environment they are considered legitimate product-
based information.
The relations between two given characteristics (either between two components or
between two attributes) are defined by links. In the case of attributes the links are defined as
dependency. In the case of links between components two requirements have to be met: (1)
Direct Influence; a change in one component is directly responsible for changing parameters in the
other, and (2) Assignment of Function; the influence has been designated by the manufacturer.
The links between the seat and legs of a chair are illustrated in Figure 1. In this example,
there are two links between the legs and the seat: Link 1, the legs support the seat; Link 2, the
seat holds the legs in place. Note also that the seat has a buckling effect on the legs. However, the
latter is not defined as a link because it does not comply with the second requirement
(assignment of function).
The complete set of links along with the internal and external components defines the
product configuration and its boundaries (see Figure 2). Note that the configuration depends
upon both the product structure and its usage context.
INSERT FIGURES 1 AND 2 HERE
Templates as Macro Operators

The configuration of a product is like a “snap shot” of the product-based information. In
order to create a new product, the dynamics of change have to be defined as well. Templates
define systematic changes between an early configuration (i.e. previous product version) and the
one that follows it (i.e. the next product version). The changes between configurations can be
expressed as combinations of elementary steps, termed operators. Figure 3 presents six
elementary (first principle) operators, which underlie the construction of the templates.2
INSERT FIGURE 3 HERE
Transition from an existing product to a new idea can be accomplished by applying these
fundamental operators in a defined sequence. These “macro” operators, when applied both to
the internal and external attributes and components, constitute the template. Note that external
attributes and components are drawn from objectively measurable factual information about the
product and its immediate environment. While the template initially draws solely on product-
based information, market-based information is subsequently examined to complete the
formulation of the product idea. In the following, we define and describe the five templates
identified in the mapping study. For ease of comprehension, all the examples analyzed pertain to
the automobile category.
The first template (described earlier, see Domino’s Pizza and glass examples), operates in
the context of product attributes. Links, in this case, pertain to dependencies between product attributes.
Thus, the Attribute Dependency Template is obtained by applying inclusion and linking operators
sequentially.
For example, on April 1, 1995, as part of a promotion campaign, the Volkswagen Motor
Company “announced” a new car model - the “Polo Harlequin” - featuring differently colored
parts. The “announcement” was not actually intended by the company, and was inserted merely
as an “April Fool’s Day” joke. However, massive response by customers calling in to place
ordersconvinced the manufacturer to go ahead and produce it. It seems that neither the
2 Note that these six operators are in fact three operators and their inverse counterparts.
customers nor the researchers could foresee the demand for such a car. The car became quite
popular in Europe and the concept of differently colored parts has even been adopted since by
some of VW’s competitors.
The concept of Polo Harlequin is consistent with the Attribute Dependency Template
and can be obtained by creating a dependency between color and specific location of a car’s parts
(e.g. door, roof). It is analogous to the already existing dependency in this category between color
and climate: Lighter colors have long been used in warmer climates, to avoid overheating and
paint damage.
Unlike the Attribute Dependency Template, which operates in the context of product
attributes, the other four templates revealed in the mapping studies operate in the context of
product components. They include the following templates: Component Control, Replacement,
Displacement, and Division. The formulation, examples and their sequence of operators are
described in Chart 1. Overall, the mapping studies conducted among several product classes
(soaps, hygiene products, bank accounts, and sneakers) indicated that despite some variation, the
five templates account for almost 70% of new product emergence3.
3 The distribution of templates in the soap study (based on brand-template matching of 1508 facings
appearing in two large department stores by two trained judges, who achieved 97% agreement, was as follows:
Attribute Dependency – 43%, Component Control – 13%, Replacement – 1.5%, Displacement – 6%, and
Division – 1.5%. Additional mapping studies were conducted on other product classes (hygiene products, bank
accounts, and sneakers), following the same procedure as in the study of the soap category. They produced the
same template typology and a similar order of proportions of template-matched product versions. In the three
categories the explanation of templates ranged as follows: Attribute Dependency - 40-68%, Component Control
- 10-30%, Replacement - (<1%)-19%, Displacement - 1-3%, and Division - 2-4%. These distributions indicate
recurrence of templates both among tangible products and services used by segments that do not necessarily
overlap.
INSERT CHART 1 HERE
The Template Approach and Ideation Research
The above taxonomy of templates can be used to approach the ideation task from a
perspective based upon product rather than market information. Such a perspective is particularly
useful given the traditional view of ideation as “ill-defined” (Simon, 1979) or “ill-structured”
(Tauber, 1972). The ill-defined nature of ideation stands in contrast to other important new
product activities (such as screening, diffusion, and various marketing assessment tools, e.g.
Narasimhan and Sen 1983, which are characterized as “well defined” because they lend
themselves to specific definitions in terms of numerical variables and to well-defined solution
plans. Most widely used ideation methods
(e.g. brainstorming and Random Stimulation) do not transform the ill-defined nature of the
problem into a defined and channeled process. Their validity has been questioned in a number of
experimental
studies concerning creativity and effectiveness (e.g. Bouchard 1969; Diehl and Stroebe 1987,
1991; Weisberg 1992). The main conclusion from these studies is that such methods lead to an
excess of ideas and analogies which actually obstruct the ideation process. It was found that the
source of satisfaction from this kind of ideation sessions is the process of ideation itself rather
than the quality of the ideas (note the “illusion of group productivity”, Paulus et. al, 1993). In
addition, Tauber (1972) has pointed out that the search for new ideas typically suffers from a lack
of synchronization between the two different activities - screening (which is well-defined) versus
ideation (which tends to be ill-defined).
Coping with the “ill-defined” nature of ideation tasks may be made easier by using an
approach which is consistent with two principles. The first, originating in cognitive psychology,
predicts that restricting the scope of an issue enhances inventive productivity. Perkins (1981) indicated that
thinking within a frame of reference requires sensitivity to the rules of the game and that, by
functioning within such a frame, one is better able to notice or recognize the unexpected. Finke,
whose research focused on the impact of visual information (e.g. Finke and Pinker, 1981, 1982),
noted:
“...restricting the ways in which creative cognitions are interpreted encourages creative exploration and discovery and further reduces the
likelihood that the person will fall back on conventional lines of thought. Restricting the categories, for example, forces people to think about
conceptual implications in more atypical ways, which tends to promote creative discovery ... and can force one to consider novel
interpretations of those concepts... (Finke, Ward, and Smith, 1992, p. 32).”
The second principle recommends the adoption of a structured ideation process which
best mimics the thinking pattern that people really follow when engaged in inventive thinking.
Introspective reports in experiments conducted by Finke, Ward and Smith (1992) indicated that
subjects often search for emergent features in the forms (e.g. images and objects), then
contemplate their functional properties, imagine themselves actually using these forms, and
finally, mentally elaborate on the context in which the forms should be found. This sequence of
events underlies the notion of function follows form. Accordingly, people are more likely to make
creative discoveries when they analyze novel forms and then assess what benefits they may
project rather than trying to create the optimal form solely on the basis of the desired benefits.
Compliance with these two principles should enhance the efficiency and productivity of
inventive thinking. It transforms the ideation process from “ill-defined” to more “well-defined”.
Some techniques specifically tailored for ideation tasks, such as morphological analysis (cf. Urban
and Hauser 1993; note also the HIT procedure, Tauber 1972), attempt to cope with the “ill-
defined” nature of ideation by manipulating both market and product-based information.

Morphological analysis initiates the ideation process by referring to currently existing rather than
imaginary product features. It calls for identifying the dimensions of the problem or the product
(e.g., type of ignition in cars, packaging in food categories etc.), listing all possible combinations
of dimensions, examining the feasibility of the alternatives, and selecting the best one. However,
one drawback of this method relative to the template approach is that it does not define specific
guidelines regarding which dimensions should be used and how they are to be combined.
Furthermore, since the meaning of dimensions is broadly defined and could potentially lead to
the generation of high order working matrices, a very large number of ideas can be generated. In
the absence of a prescribed reduction mechanism, the selection process for the best ideas may be
compromised (see the discussion of limitations in Tauber,1972).
In sum, the template approach integrates three major invention enabling perspectives.
The first stems from the proposition that several identified templates underlying the internal
dynamics of product evolution can predict new candidate products by providing guidelines for
ideation. The second is the “restricted scope” principle, which is embodied in the template
approach by channeling thinking along pre-defined inventive routes. The third perspective is the
function follows form principle which is manifested in the sequence of first proposing new
configurations of the product (which are template-based) and then, considering the benefits,
aesthetic values, and other market parameters. The initial configuration and the consideration of
consumer benefits, jointly form the product idea.
Structures resembling product templates that were developed and applied in other fields
have been valued as creative (e.g. Koestler 1964). Creativity perception may be enhanced because
these structures match certain attractors, namely, paths that the self organized mind tends to
follow (Kelso 1997) and because they preserve the original level of complexity (i.e. no new
attributes or components are introduced beyond those which exist in the product or in its
neighboring environment) while adding new utilities (Altschuler 1986). In particular, the
perception of linking previously unlinked phenomena as a creative act was first reported by
Koestler (1964). Koestler performed such linking within a multidimensional matrix of unlinked
parameters regarding other phenomena. The Attribute Dependency Template is a reduction of
his multidimensional matrix to a more simplified, two-dimensional structure, which is recurring
in the context of new products. It is therefore expected that ideas matching thAttribute
Dependency Template will be perceived as creative in the context of new product ideation.
The procedure for utilizing the dominant template – the Attribute Dependency Template - is
outlined below, followed by a report of two experiments examining its usefulness.
THE ATTRIBUTE DEPENDENCY TEMPLATE:
A PROCEDURAL FRAMEWORK
1) Construct a Dependency Matrix
Following the identification of a product’s internal and external attributes, a matrix of
attributes is constructed, the columns listing internal attributes and the rows listing both internal
attributes and external attributes. The task at this stage is to identify potential new dependencies
between attributes appearing on the two axes.
Figure 4 shows a sub-matrix of a dependency matrix constructed for the case of drinking
glasses, a category used in Experiment 2. The dependency matrix cell receives the value 0 if the
attributes are independent (e.g. color vs. sugar concentration) and a non-zero value if the column
variable depends on the row variable (e.g., capacity depends on height).
For illustrative purposes, assume that four cases of zero cells (i.e. cells that are initially
independent) can be identified. These are denoted as (a) to (d) in this sub-matrix. One case
(marked as 1) represents a natural dependency. For convenience, other zero cells are not marked
although they too might be relevant.
2) Introduce the Attribute Dependencies

The Attribute Dependency procedure is performed at this stage by introducing a
dependency between previously independent attributes. In terms of the dependency matrix this
entails changing the value of a zero cell. For example, the color of the glass could be made
dependent on the temperature of the content (case "a" in Figure 4). This dependency may be
created by introducing a color change (e.g. from green to red) at a critical temperature. This task
is best performed by individuals who belong to the operating environment or individuals with
high level of macroenvironmental knowledge (Lehmann and Winer 1994) because they will be
more efficient in acquiring new information and eliciting related knowledge (Alba and
Hutchinson 1987; see detailed review in Andrews and Smith 1996).
3) Obtain a Candidate New Idea

The objective at this stage is to scan the potential benefits of the ideas following the
“function follows form” principle. Consider the dependency created between color and
temperature (case “a”). By examining the potential benefits, at least three product ideas may be
generated. First, the threshold for color change could be made dependent on the risk of breakage
due to overheating. The resulting idea may be a glass that changes color when it reaches a certain
temperature in a microwave oven. A second idea may seek to maximize the benefit for the user
of the glass. In this case, the purpose of the color change may be to avoid overheating the
content of the glass. This might be particularly suitable in the case of use by children. Third, the
dependency may be applicable in a related product domain, for instance, a baby-bottle that
changes color when the contents reach a certain temperature.
Treatment of the marked zero cells (b-d) of the dependency matrix in Figure 4 in a
similar way to the above treatment of “case a” may produce additional new product ideas,
such as a coned-shaped glass that could be easily inserted into the sand at the beach (by
making the diameter of the glass proportional to height), or a glass with an insulation strip (by
making the heat insulation of the glass dependent on its height).
The two experiments, reported below, were designed to examine whether training in the
Attribute Dependency template improves outcomes compared to two widely used
unstructured methods - Lateral Thinking and Random Stimulation (Experiment 1) and to a
structured method – HIT (Experiment 2).
EXPERIMENT 1
Overview The experiment was composed of two phases. In the first phase the idea generation
setting was manipulated between groups: Participants were either trained in the template
approach, trained in Lateral Thinking, or trained in Random Stimulation. Following training, they
were asked to generate ideas either regarding baby ointment or mattresses. In the second phase,
the ideas were rated by senior marketing professionals. All subjects participating in idea
generation were university graduates or close to graduation. Their age range was 23-40 with a
mean age of 30. Preliminary analysis showed that the groups did not significantly differ in age,
education, or occupation.
Design and Procedure. 120 Subjects were randomly assigned into six groups generated by
crossing the training factor (template, lateral thinking, random stimulation) with the product
category factory (baby ointment, mattresses). Comprehension of the techniques was verified and
explanation ceased only upon participants’ indication that they could implement them in ideation.
The template training involved about four hours including practice tasks. Training time in the
competing methods was about two hours including practice tasks.
The efficacy of the template approach was assessed by comparing outcomes of the
template training group (hereafter the TT groups) with those of the lateral thinking training (TL)
and random stimulation training (TR) groups. The time framework for generating ideas (30
minutes) was identical in all three groups. Overall, the three groups generated 277 ideas, 132 for
baby ointment and 145 for mattresses.
The ideas generated by the TT group for the ointment category included: an ointment that
issues a specific odor upon urination by the baby; a series of ointments that differ in their
concentration of active ingredients depending on the sensitivity of the baby’s skin; and
introduction of two types of ointment - one for day use and one for night use (higher consistency
for increased protection at night by isolating the skin from the urine and lower consistency to
allow for skin breathing during the day). Among the ideas generated by the TL group were: An
ointment made of natural ingredients and an ointment with scent. The TR ideas included long
lasting ointment and a colored (replacing the traditional white) ointment.
Three senior marketing professionals were invited to participate as judges in the evaluation
procedure. All three professionals held MBA’s, had a record of at least 10 years of experience in
the marketing of consumer goods, and held high ranking marketing positions (at least vice
president or equivalent level). These professionals were asked to rate the ideas on two scales,
chosen in accordance with Finke’s (1990) suggestion to assess ideas by their originality level and
practical value. These scales are also compatible with the originality and usefulness measures
recently adopted in the context of new product design by Dahl, Chattopadhyay, and Gorn
(forthcoming). Accordingly, one scale measured originality with scale anchors (1) “Not original at
all” and (7) “Very original”. The second scale measured the overall value of the ideas. The judges
were asked to indicate whether they would recommend investment in implementing the idea on a
scale ranging from (1) “Not recommend at all” to (7) “Highly recommend”. The judges were
blind to the identity of the group members, to one another, to the notion of templates, and to the
purpose of the experiment.
Results
Table 1 displays the mean originality and value ratings for the three training groups. In
the ointment category significant differences were obtained between the three groups for both
the originality of the ideas (F(2,129)=52.13, p<.0001) and the value ratings (F(2,129)=97.46,
p<.0001).
INSERT TABLE 1 HERE
Comparison between the TT group and the TL and TR groups combined indicated that
ideas produced by the TT group were superior both in originality (t(129)=10.10, p<.001) and in
value (t(129)=13.85, p<.001). Comparison between the TL and TR groups showed no significant
differences in terms of originality (t(129)=1.34, p>.15) and value (t(129)=1.25, p>.20).
In planning the analysis it was recognized that unstructured techniques are geared to the
production of a large numbof ideas varying widely in quality and value whereas structured
methods are prescreened and more focused (Perkins 1981). Accordingly, the smallest number of
ideas produced by the TT group was selected and compared with a matching number of the
highest ranking ideas obtained in the TL and TR groups. The ideas were sorted in descending
order of rating. Based on this procedure the top 20 ideas were selected from each group. The
second panel of Table 1 summarizes the results of this analysis. Significant differences were
obtained among the three groups both when originality was the dependent measure
(F(2,57)=50.50, p<.0001) and when value served as the dependent measure (F(2,57)=69.64,
p<.0001). The comparison between the TT group and groups TL and TR indicated the
superiority of the TT group ideas both in originality (t(57)=10.04, p<.001) and in value
(t(57)=11.79, p<.001). No significant difference was obtained between TL and TR both in
originality (t(57)<1, n.s.) and in value (t(57)<1, n.s.).
This pattern of results was replicated for the mattress category (Table 1). As in the
ointment task, the analysis first focused on the raw ideas. Significant differences were observed
among the three groups both when originality and when value were the dependent measures
(F(2,142)=45.94, p<.0001 for originality, and F(2,142)=147.82, p<.0001 for value). Similarly,
dominance of the TT group over the TL and TR groups was obtained both for originality
(t(142)=9.48, p<.001) and for value (t(142)=17.17, p<.001). There was no significant difference
between the two non-template training techniques (t(142)=1.27, p>.20 for originality and
t(142)<1, n.s. for value).
The procedure comparing the best ideas was subsequently performed resulting in
significant differences for the originality ratings (F(2,57)=28.61, p<.0001) and for the value
ratings (F(2,57)=140.18, p<.0001). Similarly, dominance of the TT group over the TL and TR
groups was obtained both for originality (t(57)=7.47, p<.001) and value (t(57)=16.74, p<.001).
However, between the Lateral Thinking and Random Stimulation ideas there was no significant
difference (t(57)= 1.18, p>.20 for originality and t(57)<1, n.s. for value)4.
In addition to the advantage of the template approach as manifested both in the analysis
of the total set of ideas and the “best ideas” note that the highest ranking idea was an outcome
of the template training both in the ointment and the mattress categories.
Discussion
4 Examination of the ideas generated by all three groups both in Experiment 1 and in Experiment
2 indicated that templates-matched ideas were used almost exclusively by the template trained
individuals. Because the division between the template training and the remaining groups also
dichotomizes the use or non-use of the templates, it can be concluded that templates are
learnable and that they lead to more productive ideation.
The comparison between template training and training in competing unstructured
techniques demonstrates the added value of incorporating the template approach in ideation. It
should be noted that a major component of the costs involved in utilizing the TT technique is
the relatively longer training time required. However, the high ranking of the ideas generated by
the TT group suggests a “self screening” effect. Moreover, the fact that the TT group generated
fewer ideas suggests that the next step of new product (screening followed by concept testing)
should require reduced resources and time.
The effect due to training technique in Experiment 1 may be confounded by the required
training time that differs between the techniques and thus, difference in training time may
account for the results. In addition, Experiment 1 did not use a group of individuals who were
not trained at all prior to idea generation. Such group might serve as a benchmark for assessing
the added value of training in templates.
Experiment 2 was devised to address these issues. The technique taught to the control
group in Experiment 2 was HIT (Tauber 1972, presented above), which requires equivalent
training time and is deemed to be the closest “rival” technique to the template approach. In
addition, in Experiment 2 the template approach was compared with a “no-training” (hereafter
NT) group.
EXPERIMENT 2
Overview Experiment 2 used a different product category - drinking glasses. Like Experiment 1,
this experiment was composed of two phases. In the first phase the idea generation setting was
manipulated between groups by either training in the TT approach, training in the HIT
technique, or not training at all. In the second phase, the ideas were rated by three senior
marketing professionals.
Design and Procedure. One group consisted of 18 individuals who were specifically trained to
use the Attribute Dependency Template (the TT group). The second group was composed of 19
individuals who were trained to apply the HIT procedure (the HIT group). The training in the
TT and HIT groups lasted four hours, including practice tasks. The third group consisted of 18
individuals who received no training at all (the NT group). As in Experiment 1, the efficacy of
the template approach was assessed by comparing outcomes of the TT group with those of the
HIT and NT groups. The time framework for generating ideas (30 minutes) was the same in all
three groups. Overall, the three groups generated a total of 82 ideas.
Among the ideas generated by applying the Attribute Dependency Template were: A
drinking glass for babies with color varying according to the temperature of the milk (when the
milk is ready for drinking the color of the glass changes from red to blue); a tea cup with varying
insulation capabilities according to the temperature of the tea (when the tea is too hot the glass
enables fast cooling off, and upon reaching an optimal drinking temperature, the glass maintains
the desired heat for an extended period of time). Among the outcome ideas generated by
applying HIT were a glass that is purchased with a ready mixture of coffee and milk, a glass that
automatically cools the fluid content when it is poured in, and a glass that glows in the dark for
night feeding. The NT group generated ideas that included glasses that have exotic shapes, and a
unique combination of colors for a set.
Three senior marketing professionals were invited to participate as judges in the evaluation
procedure. Two of the judges were owners of marketing consulting agencies and one was a
senior product manager. All three judges (who did not participate in experiment 1) had at least 7
years of experience in marketing positions. The evaluation procedure and scales were identical to
those used in Experiment 1.
Results
In analyzing the total set of ideas a significant difference was obtained between the three
groups for both the originality of ideas (F(2,79)=7.5, p<.001) and their value ratings
(F(2,79)=29.8, p<.0001). The results are presented in Table 2.
INSERT TABLE 2 HERE
Comparison between the TT group and the combination of groups HIT and NT
indicated that the TT group produced superior ideas both in originality (t(79)=3.3, p<.005) and
in value (t(79)=7.6, p<.001). A comparison between the HIT and NT groups showed no
advantage in terms of originality (t(79)=1.1, n.s.) or value (t(79)<1, n.s.).
Since Tauber’s (1972) HIT does not confine the number of ideas, the more stringent
analysis performed in Experiment 1 focusing only on the better ideas was repeated in Experiment
2. Accordingly, the ideas were sorted in descending order of rating and the top 17 ideas were
selected from each group, matching the lowest number of ideas generated by the TT and NT
groups. The results are presented in the second panel of Table 2. Significant differences were
obtained among the three groups for originality (F(2,48)=5.44, p<.001) and value (F(2,48)=32.3,
p<.0001). The comparison between the TT and groups HIT and NT combined indicated that the
TT
group ideas dominated both in originality (t(48)=3.2, p<.005) and in value (t(48)=7.6, p<.001).
HIT training outperformed no training at all in value (t(48)=2.78, p<.01) although not in
originality (t(48)= (t(48)<1, n.s.). In addition to the observed superiority of the TT approach
both in the analysis of the total set of ideas and the “best ideas”, the highest ranking idea was also
“template-based”.
CONCLUSIONS AND IMPLICATIONS
The advantages of the template approach lie both in its consistency with related theories on
ideation and in its practical usefulness. On the theoretical level, the templates are obtained by
analyzing the in the evolutionary process of successful products at their mature stage and
applying them in non-mature situations. This process is analogous to “stepping back” in time. By
learning the regularities of product evolution through their internal dynamics the uncertainty
associated with innovative ideas is reduced.
The template taxonomy represents a step toward modeling the dynamics of product
ideation. If the parameters of such a model are based on product-based trends, the templates may
have the capacity to facilitate the generation of new ideas with limited external information (e.g.,
consumer needs). In addition, the template approach removes thought barriers that might hinder
ideation when traditional methods are utilized. Consider, for example, the car radio with its
antenna removed. On first examination this situation would appear inconsistent and unrealistic.
It provides no immediate cues as to how to resolve the inconsistency other than reinstating the
antenna. The template scheme opens up routes for taking advantage of and resolving such
inconsistencies.
With regard to its practical usefulness, several key advantages are worth noting. First, as
demonstrated in the mapping studies, the majority of product versions can be described by as
few as five templates. Furthermore, analysis within each of the categories supported the template
framework. Second, the learning of the procedure facilitates common understanding regarding
how to cope with the creativity task. Third, the template approach has a technological advantage
in that the ideation task begins with existing (and therefore available) products.
With regard to generalizability, two important issues need to be taken into account.
First, whereas the majority of products lend themselves to natural decomposition into
components and assignment of functions, others do not. This does not preclude the use of
the template approach but it requires consideration of other assumptions that are specific to
the product and its environment. Second, due to the simplicity and the prescribed method
sequence, individuals from heterogeneous backgrounds (e.g. different product lines,
departments, etc.) can apply the template approach and share a “common language of design”
(Mahajan, Rao, and Srivastava, 1994; Wagner and Hayashi, 1994).
REFERENCES
Alba, Joseph W. and Wesley J. Hutchinson (1987), “Dimensions of Consumer Expertise,” Journal
of Consumer Research, 13, 411-454.
Altschuler, Genrich S. (1986), To find an idea: Introduction to the theory of solving problems of inventions,
Novosibirsk, USSR, Nauka.
Altschuler Genrich S. (1985), Creativity as an Exact Science, New York, Gordon and Breach.
Andrews Jonlee, and Daniel C. Smith (1996), “In Search of the Marketing Imagination: Factors
Affecting the Creativity of Marketing Programs for Mature Products,” Journal of Marketing
Research. May, 174-187.
Boden, Margaret (1992), The Creative Mind, London: Abacus.
Bouchard Thomas J. (1969), “Personality, Problem Solving Procedure and Performance in Small
Groups,” Journal of Applied Psychology, 53, 1-29.
Chattopadhyay A. and Joseph W. Alba (1988), “The Situational Importance of Recall and
Inference in Consumer Decision Making,” Journal of Consumer Research, Vol. 15 (June), 1-12.
Cooper G. Robert (1979), “The Dimensions of Industrial New Product Success and Failure,”
Journal of Marketing, 43(3), p. 93-103.
Crawford, C. M. (1991), New Products Management. Boston, MA: Irwin.
Dahl, W. D., Amitava Chattopadhyay, and Gerald J. Gorn (Forthcoming), “The Use of Visual
Mental Imagery in New Product Design”, Journal of Marketing Research.
Dasgupta, Subrata (1994), Creativity in Invention and Design - Computational and Cognitive Explorations
of Technological Originality, Cambridge University Press.
De Bono, Edward (1992), Serious Creativity. Harper Collins Publishers.
Diehl Michael and Wolfgang Stroebe (1987), “Productivity Loss in Brainstorming Groups:
Toward the Solution of the Riddle,” Journal of Personality and Social Psychology, 53, 497-509
Diehl Michael and Wolfgang Stroebe (1991), “Productivity Loss in Idea-Generation Groups:
Tracking Down the Blocking Effects,” Journal of Personality and Social Psychology, 61, 392-403
Durgee, Jeffrey F., Gina C. O’Connor and Robert W. Veryzer (1996), “Using Mini-Concepts to
Identify Opportunities for Really New Product Functions”, MSI working paper series, Report
No. 96-105.
Finke Ronald. A. (1990), Creative Imagery: Discoveries and Inventions in Visualization. Hillsdale,
NJ:Erlbaum.
Finke Ronald.A and S. Pinker (1981) “Mapping the visual field in mental field in mental
imagery,” Journal of Experimental Psychology : General, 110, 501-517.
Finke Ronald.A and S. Pinker, (1982), “Spontaneous imagery scanning in mental extrapolation,”
Journal of Experimental Psychology: Learning, Memory, and Cognition. 8, 142-147.
Fink, Ronald A, Thomas B. Ward and Steven M. Smith (1992), Creative Cognition. Cambridge
Massachusetts: MIT Press.
Griffin, Abbie and John R . Hauser (1993), “The Voice of the Customer,” Marketing Science, 12,
63-73.
Griffin Abbie and Albert L. Page (1996), “PDMA Success Measurement
Project: Recommended Measures for Product Development Success and
Failure,” Journal of Product Innovation Management, Vol. 13, 478-96.
Klein, Arthur R. (1990), “Organizational Barriers to Creativity... and How to Knock Them
Down,” The Journal of consumer Marketing, 7, 65-66.
Kelso J.A. Scott (1997), Dynamic Patterns: The Self-Organization of Brain and Behavior. Cambridge MA:
MIT Press.
Koestler Arthur (1964), The Act of Creation, NY: McMillan.
Lehmann, Donald R., and Winer, Russel (1994), Product Management. Homewood, IL: Richard D.
Irwin.
Mahajan, Vijay, Vithala R. Rao, and Rajendra K. Srivastava (1994), “An Approach to Asses the
Importance of Brand Equity in Acquisition Decisions,” Journal of Product Innovation Management,
June, 221-235.
Marschak, T., Glennan T.K., and Summers R. (1967), Strategy for R&D Studies in Microeconomics of
Development. New York: Springer-Verlag.
Myers, James H. (1976), “Benefit Structure Analysis: A New Tool for Product Planing,” Journal of
Marketing, October 23-32.
Narasimhan, Chakravarthi and Subrata K. Sen (1983), “New Product Models For Test Market
Data,” Journal of Marketing, 47, 11-24.
Paulus Paul, Mary T..Dzindolet, George Poletes, and Mabel L. Camacho (1993), “Perception of
Performance in Group Brainstorming: The illusion of Group Productivity”, Personality and Social
Psychology Bulletin, Vol. 19, 78-89.
Perkins, David N. (1981 ),The mind’s Best work, Boston: Harvard University Press.
Petroski, Henry (1994), The Evolution of Useful Things. New York: Vintage Books.
Simon Herbert A. (1979), Models of Thoughts, New Haven: Yale University Press.
Srinivasan V., William S. Lovejoy and David Beach (1997), “Integrated Product Design for
Marketability and Manufacturing,” Journal of Marketing Research, Vol. 34, 154-63.
Tauber, Edward M. (1972), “HIT: Heuristic Ideation Technique - A Systematic Procedure for
New Product Search,” Journal of Marketing, 36, 58-61.
Ulrich Karl T. and S. Pearson (1993), “Does Product Design Really Determine 80% of
Manufacturing Cost?” Working paper 3601-93-MSA, Sloan School of Management, M.I.T.
Urban, Glen L. and John R. Hauser (1993), Design and Marketing of New Products. New Jersey:
Prentice Hall.
Von Hippel, Eric (1984), “Novel Product Concepts From Lead User: Segmenting Users By
Experience,” MSI Report No. 84-109.
Wagner, Christian and Albert Hayashi (1994), “A New Way to Create Winning Product Ideas,”
Journal of Product Innovation Management, March, 146-155.
Weisberg, Robert W. (1992), Creativity Beyond the Myth Of Genius, New York: W. H. Freeman
Company.
Wind Jerry. and Vijay Mahajan (1997), “Issues and opportunities in New Product Development:
An Introduction to the Special Issue,” Journal of Marketing Research, Vol. 34, 1-12.
Woodman, Richard W., John E. Sawyer, and Ricky W. Griffin (1993), “Towards a Theory of
Organizational Creativity,” Academy of Management Review, 18, 2, 293-321.
FIGURES, TABLES, AND CHARTS
Figure 1 : Link 1 - legs support the seat and hold it at the desired height. Link 2 - the sholds the
legs in place.
seat
2 1
legs
Figure 2: The configuration of an ordinary chair (oriented graph)
human
body
6
5
4
Seat back
2 legs
floor
Figure 3: The Operators Underlying the Templates
Operator Definition Illustration
1) Exclusion The exclusion operator

removes an unlinked
component from the c
configuration boundaries.
c
2) Inclusion The inclusion operator
introduces an external
component into the
configuration boundaries
c c
3) Unlinking An Unlinking operator
eliminates a link.
4) Linking A linking operator

connects two unlinked
components or attributes. c c
5) Splitting A splitting operator

removes an internal
c c
component from the
link. The link maintains
the original functions.
c c
6) Joining A joining operator adds a
(new) component to a
dangling link
Figure 4: Dependency sub-matrix for the case of a drinking glass.
Height Diamet Color Heat ......
of glass er
Conductivit
y
Thickness of 1
base
Height of glass b c
Temperature a
% Sugar d
....
Table 1: Experiment 1: Mean Originality and Value Ratings
Originality Value
TT TR TT TR
Trained in TL Trained in Trained in TL Trained in
Templates Trained in Random Templates Trained in Random
Lateral Stimulatio Lateral Stimulation
n
Thinking Thinking
Raw Ideas
Ointment 5.471 3.57 3.36 5.971 3.09 2.82

(0.31) (0.82) (0.93) (0.43) (0.93) (0.97)
Mattress 5.831 3.47 3.22 5.791 2.71 2.64

(0.50) (1.19) (0.51) (0.42) (0.75) (0.83)
Best Ideas
Ointment 5.471 4.27 4.22 5.971 3.75 3.68

(0.31) (0.45) (0.55) (0.43) (0.76) (0.83)
Mattress 5.831 4.86 4.66 5.791 3.27 3.21

(0.50) (0.55) (0.51) (0.42) (0.55) (0.67)
Note:
1
Represents significance (at least at p<.05 level) in the contrast between TT and the
combined TL and TR groups.
Table 2: Experiment 2: Mean Originality and Value Ratings
Originality Value
TT HIT NT TT HIT NT
Trained in Trained in No Trained in Trained in No
Templates HIT training Templates HIT training
procedure at all procedure at all
Raw Ideas
Glass 5.201 3.83 4.15 4.821 2.99 2.91

(1.02) (1.35) (1.15) (0.90) (0.99) (0.36)
Best Ideas
Glass 5.201 3.94 4.15 4.821 3.592 2.91

(1.02) (1.48) (1.15) (0.90) (0.65) (0.36)
Note:
1
Represents significance (at least at p<.05 level) in the contrast between TT and the
combined HIT and NT groups.
2
Represents significance (at least at p<.05 level) in the contrast between HIT
and NT.
Chart 1: The Other Four Revealed Templates
Template Description Example Sequence of

Operators
Component The template involves the A new electronic device connecting The Component
Control creation of a link in the form the battery of a car to the car body Control Template
of control of one internal in order to inhibit corrosion and is obtained by
component over another rust. The control is obtained by applying inclusion
internal or external providing an excess of electrons to and linking
component. the cathode, thus enabling operators,
regulation of the electrostatic sequentially.
charge, since positive charge
hinders electrochemical corrosion.
Application of this template Consider a car radio. The internal The replacement
Replacement involves the removal of an component, in this case, the car template involves
essential internal component antenna, is removed but its the sequential
from the configuration while associated intrinsic function application of the
maintaining the link between (reception of broadcast waves) is splitting, excluding,
the removed component and maintained. The resultant including, and joining
the remaining components. intermediate configuration is a operators.
This operation creates a necessary step in the replacement
temporarily inconsistent procedure even though it
abstract structure. Because of represents an incomplete product
the dangling link, the structure. The unsaturated function
operation is completed only can be fulfilled by a component
when the missing component that is external to the car radio, in
is replaced by another already this case a defroster. Finally, the
existing component: The external component is incorporated
replacement has to be an external by applying the joining operator,
component which can perform a and the configuration of a new
function similar to that provided product is obtained -- a car radio
by the one removed. that does not require an external
antenna.
Displacem An essential internal An example of the Displacement The Displacement
ent component is removed from template is excluding the car roof Template involves
the configuration. However, and its function and the new the sequential
in contrast to the product is a convertible car8. application of the
Replacement Template, its splitting, excluding,
associated link is removed as well. and unlinking
In this case, a new idea for operators.
the product has to be based
on a new appeal, one that the
former product did not
provide.
Splitting one component into Dividing a shock absorbing system The Division
Division several components which into a four-way suspension to Template involves
either contribute individually improve smooth driving and the sequential
to the accomplishment of its balance. application of the
function, or become splitting and linking
responsible for differential operators.
sub-functions.
8 Some convertible cars have a folding roof whereas in others the roof is attachable but not an integral part of the
car. The latter is a case of displacement

Toward Identifying The Inventive Templates of New Products: A Channeled Ideation Approach

Uploaded by

Copyright:

Available Formats

Toward Identifying The Inventive Templates of New Products: A Channeled Ideation Approach

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Toward Identifying The Inventive Templates of New Products: A Channeled Ideation Approach

Uploaded by

Copyright:

Available Formats

Reprinted with permission from the Journal of Marketing Research, published by the

JACOB GOLDENBERG, DAVID MAZURSKY, and SORIN SOLOMON*

TOWARD IDENTIFYING THE

enabling the individual to be more productive and focused.

immediately thereafter, it is discounted. This specific dependency could be applied in other

replicable pattern has been identified.

Templates as Product-Based Trends

products and services and, thus, it is defined as a template.

market-based information and/or product-based information, the template approach may be

The added value of product-based information is especially apparent in more complex

especially valuable when market-based information is less accessible or invalid.

when respondents describe problems in terms of current products.

One possible outcome of excessive reliance on market-based information is the

such information will be similarly effective in the context of ideation.

developed by Altschuler (1985). We examined it for suitability in the context of product

evolution. Altschuler conducted a backward analysis of problem-solution relationships and

succeeded in identifying about 200 common phenomenological patterns, which he termed

“standards”. Although Altschuler’s approach was aimed at inventive problem solving of

chronological development of products. Existing data on the evolution of soap-related products

development of the template taxonomy are provided below.

INSERT FIGURES 1 AND 2 HERE

Templates as Macro Operators

INSERT FIGURE 3 HERE

based information, market-based information is subsequently examined to complete the

the automobile category.

some of VW’s competitors.

five templates account for almost 70% of new product emergence3.

The Template Approach and Ideation Research

themselves to specific definitions in terms of numerical variables and to well-defined solution

plans. Most widely used ideation methods

ideation (which tends to be ill-defined).

defined” nature of ideation by manipulating both market and product-based information.

compromised (see the discussion of limitations in Tauber,1972).

consumer benefits, jointly form the product idea.

parameters regarding other phenomena. The Attribute Dependency Template is a reduction of

his multidimensional matrix to a more simplified, two-dimensional structure, which is recurring

outlined below, followed by a report of two experiments examining its usefulness.

THE ATTRIBUTE DEPENDENCY TEMPLATE:

1) Construct a Dependency Matrix

Following the identification of a product’s internal and external attributes, a matrix of

between attributes appearing on the two axes.

variable depends on the row variable (e.g., capacity depends on height).

although they too might be relevant.

2) Introduce the Attribute Dependencies

Hutchinson 1987; see detailed review in Andrews and Smith 1996).

3) Obtain a Candidate New Idea

changes color when the contents reach a certain temperature.

competing methods was about two hours including practice tasks.

baby ointment and 145 for mattresses.

lasting ointment and a colored (replacing the traditional white) ointment.

purpose of the experiment.

INSERT TABLE 1 HERE

differences in terms of originality (t(129)=1.34, p>.15) and value (t(129)=1.25, p>.20).

(t(57)=11.79, p<.001). No significant difference was obtained between TL and TR both in

originality (t(57)<1, n.s.) and in value (t(57)<1, n.s.).

t(142)<1, n.s. for value).

should require reduced resources and time.

the added value of training in templates.

three groups. Overall, the three groups generated a total of 82 ideas.