Feature Fatigue: When Product Capabilities Become Too Much of a Good Thing

Abstract

As technology advances, it becomes more feasible to load products with a large number of features, each of which individually might be per-ceived as useful. However, too many features can make a product over-whelming for consumers and difficult to use. Three studies examine how consumers balance their desires for capability and usability when they evaluate products and how these desires shift over time. Because con-sumers give more weight to capability and less weight to usability before use than after use, they tend to choose overly complex products that do not maximize their satisfaction when they use them, resulting in "feature fatigue." An analytical model based on these results provides additional insights into the feature fatigue effect. This model shows that choosing the number of features that maximizes initial choice results in the inclu-sion of too many features, potentially decreasing customer lifetime value. As the emphasis on future sales increases, the optimal number of fea-tures decreases. The results suggest that firms should consider having a larger number of more specialized products, each with a limited number of features, rather than loading all possible features into one product.

Full text

Journal of Marketing Research
Vol. XLII (November 2005), 431–442
431
©2005, American Marketing Association
ISSN: 0022-2437 (print), 1547-7193 (electronic)
*Debora Viana Thompson is a doctoral candidate (e-mail: dthompso@
rhsmith.umd.edu), Rebecca W. Hamilton is an assistant professor (e-mail:
rhamilto@rhsmith.umd.edu), and Roland T. Rust is David Bruce Smith
Chair in Marketing, Director of the Center for Excellence in Service, and
Chair of the Department of Marketing (e-mail: rrust@rhsmith.umd.edu),
Robert H. Smith School of Business, University of Maryland. The authors
thank Don Lehmann, Ross Rizley, Dick Wittink, and the anonymous JMR
and MSI reviewers for their helpful comments on previous drafts of this
article. They also appreciate the suggestions of participants at the 2004
Frontiers in Services Conference and the assistance of Saurabh Jain in
building the computerized experiment. This research was supported by
grants from the Marketing Science Institute and the Robert H. Smith
School of Business at the University of Maryland.
DEBORA VIANA THOMPSON, REBECCA W. HAMILTON, and ROLAND T. RUST*
As technology advances, it becomes more feasible to load products
with a large number of features, each of which individually might be per-
ceived as useful. However, too many features can make a product over-
whelming for consumers and difficult to use. Three studies examine how
consumers balance their desires for capability and usability when they
evaluate products and how these desires shift over time. Because con-
sumers give more weight to capability and less weight to usability before
use than after use, they tend to choose overly complex products that do
not maximize their satisfaction when they use them, resulting in “feature
fatigue.” An analytical model based on these results provides additional
insights into the feature fatigue effect. This model shows that choosing
the number of features that maximizes initial choice results in the inclu-
sion of too many features, potentially decreasing customer lifetime value.
As the emphasis on future sales increases, the optimal number of fea-
tures decreases. The results suggest that firms should consider having a
larger number of more specialized products, each with a limited number
of features, rather than loading all possible features into one product.
Feature Fatigue: When Product Capabilities
Become Too Much of a Good Thing
A common way to enhance and differentiate a product is
by increasing its number of features (Goldenberg et al.
2003; Mukherjee and Hoyer 2001; Nowlis and Simonson
1996), which provides greater functionality for consumers.
This strategy has become especially popular as new devel-
opments in electronics and information technology (e.g.,
miniaturization and integration of electronic components)
have enabled products to include more functions yet cost
less and require less time to be manufactured (Freund,
König, and Roth 1997).
Each additional feature provides another reason for the
consumer to purchase a product (Brown and Carpenter
2000) and may add desired capabilities, but too many fea-
tures can make products overwhelming for consumers,
leading to dissatisfaction and “feature fatigue.” Anecdotal
evidence suggests that consumers do not use all the features
of the products they buy (Ammirati 2003), and even more
significantly, empirical evidence indicates that consumers
may experience negative emotional reactions, such as anxi-
ety or stress in response to product complexity (Mick and
Fournier 1998).
Why do consumers seem to make choices that do not
maximize their long-term satisfaction? One potential reason
is that consumers do not make a connection between increas-
ing the number of product features and the difficulty of using
aproduct. Another reason is that consumers understand that
products with more features will be more difficult to use, but
because features are bundled together, they are forced to buy
features they do not want in order to obtain features they do
want. Finally, consumers may understand that products with
more features will be more difficult to use, but they may give
ease of use too little weight in their purchase decisions.
In this research, we examine how consumers balance
their competing needs for functionality and ease of use
when evaluating products. First, we measure the effects of
adding product features on two distinct product dimensions,
the perceived capability of the product and the perceived
usability of the product. Second, we test the degree to
which consumers consider usability compared with capabil-
ity when evaluating products before using them. Third, we
measure the relative weights of capability and usability in

432 JOURNAL OF MARKETING RESEARCH, NOVEMBER 2005
consumers’ expected utility (before use) and experienced
utility (after use) and test for significant differences in these
weights before and after product use. Whereas previous
research has focused on either preusage evaluations, such as
purchase intentions (e.g., Carpenter, Glazer, and Nakamoto
1994), or postusage evaluations, such as satisfaction (e.g.,
Bolton and Lemon 1999) and usability (e.g., McLaughlin
and Skinner 2000), we integrate these perspectives by com-
paring evaluations of products both before and after use.
We organize the article as follows: First, we briefly dis-
cuss the effects of adding product features on consumers’
evaluations of products. Second, we report the results of
three studies we designed to test our hypotheses. On the
basis of our results, we propose an analytical model to help
managers balance the sales benefits of adding features
against the customer equity costs of feature fatigue. We
conclude with a discussion of our results, their theoretical
and managerial implications, and directions for further
research.
THE EFFECTS OF ADDING PRODUCT FEATURES ON
PRODUCT EVALUATIONS
Both economic theory and current market research tech-
niques predict that increasing the number of features will
make products more appealing. Economic theory models
consumers’ preferences using an additive utility function
that links product attributes to consumer demand (Lancaster
1971). Each positively valued attribute increases con-
sumers’ utility. Similarly, market research techniques, such
as conjoint analysis or discrete choice analysis, model each
product as a bundle of attributes and estimate partworths for
each attribute (Srinivasan, Lovejoy, and Beach 1997).
Because market shares are predicted on the basis of these
partworths, each positively valued feature increases a prod-
uct’s market share compared with products without the
feature.
The behavioral assumption underlying decompositional
models such as these is that consumers infer functional
product benefits from concrete product attributes. Because
the utility of a product is based on its potential benefits to
the consumer rather than product features per se, consumers
translate information about concrete product attributes into
functional benefits in their mental representations (Olson
and Reynolds 1983). Consistent with this mapping process,
research has shown that added features provide positive dif-
ferentiation by giving a product perceived advantages over
competitive products (Carpenter, Glazer, and Nakamoto
1994). Consumers seem to use added features in an instru-
mental reasoning process that makes the brand with more
features appear superior in a choice set (Brown and Carpen-
ter 2000). Although these inferences have been demon-
strated to occur for both irrelevant and important attributes
(Brown and Carpenter 2000), consumers must perceive a
benefit from the added feature for product evaluations to
increase. Nonnegative features that consumers perceive to
add little or no value (e.g., calculator functions that are use-
ful only to biochemistry students) tend to decrease brand
share because they provide reasons against choosing the
enhanced product (Simonson, Carmon, and O’Curry 1994).
Thus, we predict that perceived product capability (i.e.,
the consumer’s beliefs about the product’s ability to per-
form desired functions) will increase as more features that
provide perceived benefits are added to a product. Whereas
previous research has asked participants to compare prod-
ucts that differ on a single feature (e.g., Brown and Carpen-
ter 2000), we predict that consumers will perceive greater
capability as the number of features increases, even when
evaluating a single product. Moreover, whereas previous
research has focused on consumer perceptions before use,
we predict that this relationship will hold both before and
after product use.
H1: As the number of beneficial features included in a product
increases, perceptions of the product’s capability increase.
In addition to the product’s capability, consumers should
consider their ability to use the product and benefit from its
features. Research on usability and user-centered design
suggests that adding features to products has a negative
effect on consumers’ ability to use them across several
product categories (Wiklund 1994). Every additional fea-
ture is “one more thing to learn, one more thing to possibly
misunderstand, and one more thing to search through when
looking for the thing you want” (Nielsen 1993, p. 155).
Usability research has focused on measures that enable a
consumer’s usage experience to be compared across prod-
ucts, such as the ease of learning how to use a product, the
propensity to make errors while using it, and the efficiency
of using it (McLaughlin and Skinner 2000). The time taken
to complete a task, the ratio of successful to unsuccessful
interactions with a product, and the number of errors are
typical operationalizations of usability (Nielsen 1993).
However, although usability research supports the principle
that less is more, this research is based on consumers’ actual
experiences using products rather than their perceptions
about their ability to use products.
There is some evidence that consumers account for learn-
ing costs when features are added to products. For example,
adding a novel feature to a Web television or personal com-
puter had a positive effect on product evaluations when the
feature was described as fully automatic, but it had a nega-
tive effect on product evaluations when it was described as
manually operated, presumably due to consumers’ infer-
ences about learning costs (Mukherjee and Hoyer 2001).
However, although these findings are suggestive, consumer
perceptions were measured in response to varying a single
feature across products, and consumers did not use the
products being evaluated.
On the basis of both usability studies and consumers’
inferences about the effects of adding a feature to a product,
we predict that perceived product usability, or the con-
sumer’s beliefs about the difficulty of learning and using the
product, will decrease as more individually beneficial fea-
tures are added to a product. This should be true even when
consumers evaluate a single product and should hold both
before and after consumers use the product.
H2: As the number of beneficial features included in a product
increases, perceptions of the product’s usability decrease.
How will consumers’ expertise within a product category
affect their perceptions of product capability and product
usability? Experts have a better understanding of product-
related information and are better able to discriminate
between important and unimportant features than novices
(Alba and Hutchinson 1987). As a result, experts should be
better able to assess product capability than novices. How-
ever, whether experts perceive a given product’s capability

Feature Fatigue 433
to be higher or lower than novices will depend on the spe-
cific features of the product and the benefits they are
believed to provide. Therefore, we cannot make a general
prediction about the effect of expertise on perceived product
capability. In contrast, the effect of expertise on perceived
usability is clear. Experts perform product-related tasks
more automatically, freeing cognitive resources that can be
used to learn new product features (Alba and Hutchinson
1987). For example, experts were more successful in solv-
ing tasks and were more efficient when using a mobile
phone than novices (Ziefle 2002). Experts also may be bet-
ter able to handle complex products because they focus
their attention on a smaller, more diagnostic number of
inputs (Spence and Brucks 1997). Thus, we predict that
because experts are better able to learn and use each prod-
uct feature than novices, usability ratings will be higher for
experts than for novices.
H3: Expertise has a positive effect on perceptions of product
usability.
HOW CONSUMERS WEIGH CAPABILITY AND
USABILITY IN THEIR PRODUCT EVALUATIONS
If increasing the number of product features has positive
effects on perceived capability (H1) and negative effects on
perceived usability (H2), how do consumers integrate these
two product dimensions when forming their overall product
evaluations? Previous research suggests that consumers
consider both the benefits and the costs of adding a new
feature to a product (Mukherjee and Hoyer 2001). We pro-
pose that the net effect of increasing the number of product
features on product utility depends on the relative weights
that consumers give to capability and usability in their prod-
uct evaluations and that these weights may vary across time
and situations.
Experimental research has shown that when people eval-
uate options for the distant future, they favor highly desir-
able options that are less feasible over less desirable options
that are highly feasible. However, the opposite is true when
people evaluate options in the near future (Liberman and
Trope 1998). The relative weights of desirability (i.e., the
expected value of the goal, or the “why” aspect of an
action) and feasibility (i.e., beliefs about the difficulty of
reaching the end state, or the “how” aspect of an action)
change because the construal of more distant future events
tends to be more abstract, favoring desirability, whereas the
construal of near future events tends to be more concrete,
favoring feasibility (Liberman and Trope 1998).
Thus, we propose that consumers will create more
abstract construals of products in their evaluations before
use, assigning greater weight to the desirability of the prom-
ised benefits (e.g., What can this product do for me?), than
in their evaluations after use. In contrast, we expect that
consumers will develop a more concrete construal of the
product in their evaluations after use, placing more weight
on feasibility (e.g., Is this product easy to use?), than in
their evaluations before use. On the basis of this expected
shift in the importance of capability and usability, we pre-
dict the following:
H4: Consumers give more weight to product capability in their
expected product utilities (before use) than in their experi-
enced product utilities (after use).
H5: Consumers give less weight to product usability in their
expected product utilities (before use) than in their experi-
enced product utilities (after use).
To test our hypotheses, we ran three studies in which par-
ticipants evaluated or used Web-based products. Studies 1
and 2 examine consumers’ intuitions about the effects of
adding product features on capability (H1) and usability (H2
and H3) before use. Study 3 directly compares consumers’
ratings of capability and usability and their overall product
evaluations before and after using products (H4and H5).
Our goal is to demonstrate that though the effects of
increasing the number of features on perceptions of product
capability and usability are significant both before and after
product use, there is a shift in the relative weights of these
dimensions on consumers’ product evaluations. Figure 1
summarizes our hypotheses.
STUDY 1: CONSUMERS’ INTUITIONS
We designed Study 1 to simulate an in-store experience.
Our goal was to test how consumers’ intuitions about prod-
uct capability and usability were related to the number of
product features (H1and H2) and whether perceived usabil-
ity was related to expertise (H3).
Consumers’ involvement in the evaluation task may
affect their motivation to process product information (Celsi
and Olson 1988). For example, highly involved consumers
are more likely to elaborate on product information and
form inferences (Celsi and Olson 1988). Thus, involvement
with the task could potentially affect participants’ judg-
ments about product capability and usability. To control for
this, we manipulated involvement across conditions.
Figure 1
CONCEPTUAL MODEL
A: Before Use
Features
Capability
Usability
Expertise
Expected
utility
H1H4
H5
H2
H3
Features
Capability
Usability
Expertise
H1H4
H5
H2
H3
Experienced
utility
B: After Use

434 JOURNAL OF MARKETING RESEARCH, NOVEMBER 2005
Stimuli
To develop the stimuli for our studies, we conducted a
pretest in which 40 participants (69% females, Mage = 21.8)
rated the importance of and their familiarity with 30 fea-
tures of the following four products: a digital audio player, a
digital video player, a personal digital assistant, and an
online product-rating database. Participants also rated their
involvement and expertise with each product category. We
selected digital audio players and digital video players
because participants were involved and familiar with these
product categories. Three models of each product were cre-
ated, differing only in their number of features. The low
level of features included the 7 most important features, the
medium level included the 14 most important features, and
the high level included the 21 most important features.
Participants, Design, and Procedures
Our study comprised 130 undergraduate students (50.8%
females, Mage = 20.5) who we randomly assigned to condi-
tions. The study had a 2 (player: video, audio) ×3 (feature:
low, medium, high) ×2 (involvement: low, high) mixed
design. We manipulated player and involvement between
subjects and number of features within subjects. In the
high-involvement condition, we told participants that after
they evaluated three models, they would choose one model
to perform a series of tasks. We told low-involvement par-
ticipants simply that they would evaluate three models of
video (audio) players. We conducted the study using Media-
Lab software, and we ran sessions in a computer lab with
groups of 3 to 18 students. Participants worked individually.
Participants first rated their expertise with digital video
(audio) players. Next, they viewed the user interface and the
list of features for each model. Participants rated their per-
ceptions of each model’s capability and usability and then
provided an overall evaluation of each model. The order in
which participants evaluated the low-, medium-, and high-
feature models was counterbalanced between subjects,
according to a standard self-conjugate Latin square. After
rating all three models, we asked participants to choose one
of the models.
Measures
We measured participants’ expertise using five items
(e.g., How familiar are you with digital video [audio] play-
ers? How frequently do you watch videos [listen to music]
on your computer? Mitchell and Dacin 1996). We measured
product capability using three items (extent to which the
products were likely to perform poorly/well, offer few/a lot
of advantages, and add little/a lot of value; Mukherjee and
Hoyer 2001). We measured product usability using eight
items (e.g., Learning to use this product will be easy for me,
Interacting with this product will not require a lot of my
mental effort, It will be easy to get this product to do what I
want it to do; Chin, Diehl, and Norman 1988). We measured
expected product utility using six items (bad/good, unlik-
able/likable, not useful/useful, low/high quality, undesir-
able/desirable, unfavorable/favorable; Peracchio and Tybout
1996). After choosing one of the models, participants rated
their decision confidence and the difficulty of the choice.
All items used seven-point scales.
1The comparative fit indexes (CFI) ranged from .91 to .93, capability
items loadings ranged from .58 to .95, usability items loadings ranged
from .70 to .96, and overall evaluation items ranged from .49 to .96. Each
of the three factors had an average extracted variance greater than 62%.
Capability and usability were not correlated for any of the models. How-
ever, capability and overall evaluations were correlated for all three models
(rlow = .76, rmedium = .83, and rhigh = .69; all ps< .001), and usability and
overall evaluations were correlated for the high-feature model (r = .29, p<
.001).
Results
Reliability for expertise, capability, usability, and overall
product evaluations all exceeded .83. To assess the con-
struct validity of our capability, usability, and overall prod-
uct evaluation scales, we ran a confirmatory factor analysis
for each of the low-, medium-, and high-feature models. A
three-factor model indicated an acceptable goodness of fit
and significant loadings for each observed variable in their
respective latent factor (all ps< .001).1Involvement did not
affect any dependent measures (all ps> .13), and we col-
lapsed the data across involvement conditions.
To test H1, we ran a 2 (player) ×3 (features) repeated-
measures analysis of covariance (ANCOVA) on product
capability with expertise as a covariate. There was a main
effect of number of features (F(2, 250) = 24.1, p< .001). No
other effects were significant (ps> .08). As we predicted,
the within-subjects linear contrast for capability across fea-
ture levels was significant (Flinear(1, 125) = 27.8, p< .001),
indicating that perceptions of product capability signifi-
cantly increased as the number of product features
increased (Mlow = 3.4, Mmedium = 4.9, and Mhigh = 6.0).
To test H2, we ran a 2 (player) ×3 (features) repeated-
measures ANCOVA on product usability with expertise as a
covariate. There was a significant main effect of number of
features (F(2, 250) = 17.6, p< .001). The main effect of
player and the interaction between number of features and
player were not significant (ps> .09). As we predicted, the
within-subjects linear contrast for usability across feature
levels was significant (Flinear(1, 125) = 22.7, p< .001), indi-
cating that perceptions of product usability significantly
decreased as the number of features increased (Mlow = 6.2,
Mmedium = 5.6, and Mhigh = 4.8). When we controlled for
the number of features, expertise had a positive effect on
usability (F(1, 125) = 43.1, p< .001). Perceived usability
for both video and audio players was higher for experts than
for novices, in support of H3.
A 2 (player) ×3 (features) repeated-measures ANCOVA
on product expected utility with expertise as a covariate
revealed only a significant main effect of features
(F(2, 250) = 7.5, p= .01). No other effects were significant
(ps> .16). The within-subjects linear contrast for product
expected utility across feature levels was significant
(Flinear(1, 125) = 8.4, p< .01), indicating that expected util-
ity increased as the number of features increased (Mlow =
4.1, Mmedium = 5.1, and Mhigh = 5.6). Regardless of expert-
ise, expected utility was most favorable when the product
included the highest number of features. Thus, before use,
capability appears to have a stronger effect than usability on
product expected utility. Figure 2 shows the impact of
increasing the number of features on ratings of capability,
usability, and expected utility for the video player.
After evaluating the three models, participants chose one
of them to perform a series of tasks. Participants’ choices

Feature Fatigue 435
1
2
3
4
5
6
7
Low Medium High
Capability Usability Expected utility
Number of Features
Figure 2
EFFECTS OF NUMBER OF FEATURES ON DEPENDENT
VARIABLES FOR THE DIGITAL VIDEO PLAYER (STUDY 1)
2This result is based on a median split on the expertise variable
(median = 4.0).
3In the follow-up study (N = 73), we showed participants only one
model of the video player (either low or high feature) and asked them to
provide product evaluations. The results were consistent with those of
Study 1. Perceived capability increased with number of features
(F(1, 71) = 23.8, p< .001), perceived usability decreased with number of
features (F(1, 69) = 3.9, p= .05), and expected utility increased with num-
ber of features (F(1, 69) = 8.2, p< .01).
strongly indicated a preference for products with a higher
number of features and greater capability, regardless of
expertise. The majority of the respondents chose the model
with the highest number of features (62.3%) rather than the
model with a medium number of features (28.5%) or the
model with the lowest number of features (9.2%). A multi-
nomial logistic regression of player and expertise on choice
showed that neither of these factors affected choice (all ps>
.55). Notably, despite the lack of difference in their choices,
novices rated the difficulty of choosing marginally higher
than experts (F(1, 128) = 3.5, p= .06), and experts were
more confident in their choices than novices (F(1, 128) =
9.8, p< .01).2
Discussion
The results of Study 1 suggest that consumers believe
that increasing the number of features decreases the usabil-
ity of products and increases their capability. However,
regardless of participants’ expertise, their expected product
utility and choices still favored products with a higher level
of features. Therefore, consumers’ initial preferences appear
to be driven more by product capability ratings than by
usability ratings.
One limitation of Study 1 is that varying the number of
features within subjects may have increased the salience of
the number of features when they judged capability and
usability. However, a replication of Study 1 using a
between-subjects design produced the same results, indicat-
ing that salience does not explain the effect.3We also
4The CFI ranged from .95 to .96. Capability loadings ranged from .66 to
.83 (average extracted variance was greater than 54%). Usability loadings
ranged from .46 to .92 (average extracted variance was greater than 64%).
The correlation between the two factors was not significant.
address this concern by using a between-subjects design in
Study 2. A second limitation of Study 1 is that because the
three models of video and audio players were the same for
all participants, they may have included features that par-
ticipants did not consider important, potentially decreasing
usability without adding significant capability. Although
this is a realistic choice situation—companies often find it
cheaper to produce feature-rich products that can satisfy the
needs of heterogeneous consumers than to produce more
narrowly targeted products with fewer features—we want to
disentangle supply side and demand side explanations for
feature fatigue. In Study 2, we allow participants to cus-
tomize their products so that the products being evaluated
include only desired features.
STUDY 2: CUSTOMIZING A PRODUCT
In Study 2, participants customized their own products
by selecting the features they would like to add from a list
of features. We predicted that consumers who chose more
features would perceive their products to have more capa-
bility but less usability than consumers who chose fewer
features. Support for H2will show that consumers predict
degradation in usability as the number of features increases,
even when products include only desirable features.
Participants, Design, and Procedures
Study 2 comprised 141 undergraduate students (55.3%
females, Mage = 21.1). We asked the participants to imagine
that they were about to subscribe to and download a new
digital audio player and a digital video player and that they
would have the opportunity to choose the features they
wanted. Product category was manipulated within subjects.
The order in which they designed the two products was
counterbalanced between subjects.
As in Study 1, we used a digital audio player and a digi-
tal video player as our products. For each product, we pre-
sented participants with 25 different features that they could
select. Participants checked off each feature they wanted to
include in the product they were buying. To isolate the
effects of usability constraints from the effect of financial
constraints, we informed them that their budget for the pur-
chase would allow them to select as many features as they
wanted. After selecting features, participants rated the prod-
uct’s perceived capability and usability. We measured prod-
uct capability, product usability, and expertise using the
same scales as in Study 1. Participants also rated their
familiarity with each feature and the importance of each
feature (1 = “not at all important/familiar” and 7 = “very
important/familiar”).
Results
The reliability for expertise, capability, and usability
ranged from .78 to .93. A confirmatory factor analysis on
the capability and usability measures for each media player
supported the construct validity of these constructs. A two-
factor solution yielded a reasonable goodness of fit and sig-
nificant loadings of each observed variable in their respec-
tive factor (all ps< .001).4The order in which participants

436 JOURNAL OF MARKETING RESEARCH, NOVEMBER 2005
customized the products was not correlated with any of our
measures (all ps> .10), except with usability for the video
player (p= .04). We included order as a covariate in all
analyses related to the perceived usability of the video
player.
The average number of features chosen among the 25
available was 19.6 (standard deviation = 4.8) for the video
player and 19.6 (standard deviation = 4.3) for the audio
player. Approximately half of the sample chose more than
80% of the available product features, and the median num-
ber of features chosen for both players was 20. Notably,
although the specific features chosen by experts and
novices differed, the number of features chosen by experts
and novices did not differ (ps> .25). Experts reported sig-
nificantly greater familiarity with all 25 video player fea-
tures and with 23 of the 25 audio player features. The fea-
tures that experts chose more frequently were among those
rated least familiar by novices. For example, the three audio
player features that experts chose significantly more fre-
quently than novices (i.e., the equalizer/bass boost, preamp
and equalizer settings, and encoded file name control) were
three of the seven features for which the difference in famil-
iarity ratings between experts and novices was largest.
In H1,wepredicted that participants who chose more fea-
tures would perceive that their products had greater capability
than participants who chose fewer features. As we expected,
when we regressed ratings of product capability on the num-
ber of selected features and expertise, we found a positive and
significant coefficient for both the video player (β=.50, t =
6.9, p<.001) and the audio player (β=.47, t = 6.2, p<.001),
in support of H1.The effect of expertise on capability was not
significant for either the video or the audio player (ps>.07).
We predicted that usability would have a negative rela-
tionship with number of features (H2)and a positive relation-
ship with expertise (H3). We found a significant, negative
effect of number of selected features on the perceived usabil-
ity of the video player (β=–.16, t = –2.2, p=.03). However,
the effect was not significant for the audio player (β=.01,
t=.70, p=.48). Thus, the findings partially support H2.
When we controlled for the number of features, expertise
had a significant, positive effect on perceived usability for
both players (video player: β=.52, t = 7.0, p<.001; audio
player: β=.98, t = 52.9, p<.001), in support of H3.
Discussion
Overall, the results of Study 2 support our predictions.
The number of features participants selected increased per-
ceived product capability for both products and decreased
perceived product usability for one of the two products.
Thus, the connection between adding product features and
decreasing usability seems to hold even when the consumer
individually selects each of the included features. Consis-
tent with our expectations, expertise significantly improved
ratings of product usability but did not affect ratings of
product capability.
On average, participants chose a high number of features,
again suggesting that a desire for capability is driving deci-
sions more than a desire for usability. Notably, the average
number of features chosen in Study 2 was nearly the same
as the number of features in Study 1’s high-feature condi-
tion. Using two different types of choice tasks, participants
clearly favored high-feature products over low-feature prod-
ucts. However, Studies 1 and 2 test choices before using
products. In Study 3, we compare the ratings of participants
who have not used the product with ratings of participants
who have used the product.
STUDY 3: CONTRASTING EVALUATIONS BEFORE AND
AFTER PRODUCT USE
In Study 3, we compared consumers’ evaluations of prod-
ucts with a low, medium, or high number of features before
use and after use. We expected that consumers would give
more weight to capability before use than after use (H4) and
that consumers would give less weight to usability before
use than after use (H5).
Participants, Design, and Procedures
Study 3 comprised 190 participants (52.1% males,
Mage = 20.5) who we randomly assigned to conditions using
a 2 (product use: before, after) ×2 (feature: low, high)
between-subjects design. We conducted the study using
MediaLab software, and we ran sessions in a computer lab
with groups of 2 to 18 students. Participants worked indi-
vidually. Each participant evaluated one model of the prod-
uct, either before or after product use. The use of a
between-subjects design was critical because making pre-
dictions about capability or usability before use can bias
participants’ evaluations of the product after use (Jones
1977).
The product we used in this study was the same digital
video player that participants evaluated in Study 1. Two
working models of the product were created, one with 7
features (low-features condition) and one with 21 features
(high-features condition). Participants who used the product
were provided with a manual of the video player that
described the features of their model and how to use them.
In the low-features condition, the manual had four pages,
and in the high-features condition, the manual had eight
pages. The layout of the manual was identical across
conditions.
We asked participants to imagine that they were consid-
ering subscribing to and downloading a new digital video
player. In the before-use condition, participants viewed the
user interfaces and a list of features for three models of
players, one with a low number of features (7 features), one
with a medium number of features (14 features), and one
with a high number of features (21 features). The order of
presentation was counterbalanced between subjects. Partici-
pants evaluated either the low- or the high-feature model
and then chose their preferred model.
We told participants in the after-product-use condition
that they would use one model of a new digital video player.
We asked them to perform a series of four tasks using either
the low- or the high-feature model of the player. These tasks
included choosing a specific movie from a playlist, watch-
ing parts of the movie, modifying the audio settings, and
recording parts of another movie available in the playlist.
After completing these tasks, participants were free to use
the player at their leisure. Next, participants evaluated the
product they used. After completing their evaluations, they
viewed the user interfaces and a list of features for two
additional models of digital video players (e.g., models with
a low and medium number of features if they had used the
high-features model). The order of presenting the other two

Feature Fatigue 437
5The CFI was .93. Capability loadings ranged from .94 to .97 (average
extracted variance = 91%). Usability loadings ranged from .52 to .93 (aver-
age extracted variance = 64%), and overall evaluation loadings ranged
from .55 to .92 (average extracted variance = 70%). The correlation
between usability and capability was not significant. Overall evaluations
were correlated with capability (r = .63, p< .001) and usability (r = .29,
p< .001).
models was counterbalanced between subjects. Finally, par-
ticipants chose their preferred model.
Measures
We measured expertise and product usability using the
same measures as in Studies 1 and 2. We measured product
capability using three items (this digital video player per-
forms many functions, has many capabilities, and has a
large number of features). We measured expected and expe-
rienced utilities separately using the six-item measure for
overall product evaluation that we used in Study 1 and one
item about product satisfaction (How satisfied would you be
if you subscribed to the digital player? [in the before-use
condition] How satisfied were you with the digital player
you used? [in the after-use condition]). We measured all
items using seven-point scales.
After participants had either evaluated or used one of the
models, we asked them to choose one of the three models.
As in Study 1, participants rated their confidence in their
decision and the difficulty of making the decision. We also
recorded participants’ clickstreams as they used the video
player in the after-use condition. We gathered information
on how many tasks participants completed, the time it took
them to complete the tasks, and how long they used the
player.
Results
The reliability of the multiple-item scales ranged from
.89 to .98. A confirmatory factor analysis on the capability,
usability, and overall product evaluation scales showed an
acceptable goodness of fit for the three-factor solution and
significant loadings for each observable variable in their
respective latent factors (all ps< .001).5Because order was
not significant for any of the dependent variables (all ps>
.06), we collapsed the data across order conditions for sub-
sequent analyses. Table 1 shows the means of the dependent
variables across conditions.
A 2 (product use) ×2 (features) ANCOVA on perceived
capability with expertise as a covariate showed a significant
6We also estimated the models using partial least squares, and the results
were consistent with those we obtained using maximum likelihood
estimation.
main effect of number of features (F(1, 185) = 132.9, p<
.001), indicating that capability increased with the number
of features (Mlow = 3.2, Mhigh = 5.2). Thus, there is support
for H1. In addition, we found a significant main effect of
product use (F(1, 185) = 5.2, p= .02). Perceived product
capability was lower after use (Mafter = 4.0) than before use
(Mbefore = 4.4). The interaction between number of features
and product use on ratings of capability was also significant
(F(1, 185) = 67.2, p< .001), indicating that the number of
features had a smaller effect on perceptions of product
capability after use than before use. The effect of expertise
on perceived product capability was not significant (p>
.60).
A 2 (product use) ×2 (features) ANCOVA on perceived
usability with expertise as a covariate showed that usability
significantly decreased with the number of features
(F(1, 185) = 33.1, p< .001), in support of H2(Mlow = 5.9,
Mhigh = 4.9). Consistent with H3, participants’ expertise had
a positive effect on their perceptions of product usability
(F(1, 185) = 12.7, p< .001). No other effects were signifi-
cant (ps> .17).
Consistent with H4and H5, a 2 (product use) ×2 (fea-
tures) ANCOVA on overall product evaluations with expert-
ise as a covariate revealed a main effect of features (Mlow =
4.5, Mhigh = 5.1; F(1, 185) = 15.8, p< .001) that was quali-
fied by a significant interaction between features and prod-
uct use (F(1, 185) = 31.5, p< .001). When we controlled for
expertise, product evaluations before use significantly
increased with number of features (Mlow = 4.0, Mhigh = 5.6;
F(1, 91) = 49.0, p< .001), but product evaluations after use
did not (Mlow = 5.0, Mhigh = 4.7; F(1, 93) = 1.6, p= .20).
The effect of expertise on participants’ overall product eval-
uations was not significant (p> .40). A 2 ×2 ANCOVA on
product satisfaction produced similar results.
To investigate the relative weights of product capability
and usability on consumers’ product utilities before and
after product use, we ran a multisample path analysis using
maximum likelihood estimation.6Number of features,
expertise, and their interaction entered the model as inde-
pendent variables. We partialled out the main effects of
number of features and expertise from the interaction effect
and used the regression unstandardized residuals as the
Ta b le 1
EFFECT OF NUMBER OF FEATURES ON DEPENDENT VARIABLES (STUDY 3)
Product Use Number of Features Product Capability Product Usability Product Evaluations Product Satisfaction
Before Low 2.7a6.0a4.0a3.4a
(1.2) (.9) (1.2) (1.6)
High 6.2b5.1b5.6b5.6b
(.9) (1.1) (1.0) (1.0)
After Low 3.7c5.8a5.0c5.2b,c
(1.3) (1.1) (1.0) (1.1)
High 4.3d4.8b4.7c4.9c
(1.3) (1.4) (1.3) (1.6)
Notes: N = 190 participants. Standard deviations are in parentheses. Different superscripts in the same column indicate that the difference between means
is significant (p< .05).

438 JOURNAL OF MARKETING RESEARCH, NOVEMBER 2005
Ta b le 2
STANDARDIZED PATH COEFFICIENTS (STUDY 3)
Dependent Variables
Capability Usability Overall Evaluations Satisfaction
Before Use After Use Before Use After Use Before Use After Use Before Use After Use
Number of features .85** .20* –.43** –.36** — — — —
Expertise n.s. n.s. .36** n.s. — — — —
Capability — — — — .82** .45** .79** .48**
Usability — — — — .24** .51** .15**.51**
*p< .05.
**p< .001.
Notes: n.s. = not significant.
Independent Variables
7The CFI was .95 in the before-use sample and .99 in the after-use sam-
ple. All χ2tests > .07.
8We computed the significance level of all indirect effects using the
Sobel t statistic.
interaction term. Product capability and usability were
mediator variables. We estimated the coefficients with two
different dependent variables that reflected product utility:
overall product evaluations and satisfaction. All goodness-
of-fit indexes were in an acceptable range.7The interaction
between number of features and expertise was not signifi-
cant (ps> .11). Table 2 shows the standardized path coeffi-
cients before and after product use.
To test the difference in the relative weights of capability
and usability on expected and experienced product utility
(H4and H5), we constrained each of these two parameters
in the model to be equal across conditions and assessed
whether the chi-square decrease in the unconstrained model
was significant. The Lagrange-multiplier test showed that
the effects of product capability on product evaluations and
satisfaction differed significantly in the before-use and
after-use conditions (χ2(1)overall evaluations = 4.2, p< .01;
χ2(1)satisfaction = 4.9, p= .03). Consistent with H4, partici-
pants gave more weight to product capability before use
than after use.
In H5, we predicted that consumers would give less
weight to usability before use than after use. Our model
comparisons partially support this prediction. The effect of
usability on overall product evaluations was invariant before
and after use (p> .26), but the effect of usability on satis-
faction was significantly lower before product use than after
product use (χ2(1) = 4.5, p= .03). This indicates that, as we
expected, participants gave less weight to usability in their
predicted product satisfaction than in their satisfaction rat-
ings after using the product.
Additional Analyses
Decomposing the direct and indirect effects in our model,
we found that before product use, the indirect effect of
product features on overall product evaluations mediated by
product capability was strong (β= .70, p< .001) and over-
shadowed the significant, negative indirect effect of product
features through usability (β= –.10, p< .01), yielding a
positive net effect.8After product use, this pattern reversed.
The indirect effect of features through capability became
nonsignificant (β= .09, p> .05), and the indirect effect of
features through usability was negative and significant (β=
–.18, p= .001), resulting in a negative net effect. The indi-
rect effects of number of product features on satisfaction
followed the same pattern.
Participants’ choices of players before and after product
use show a substantial decrease in the share of the high-
feature model. The majority of the respondents in the
before-use condition (66%) chose the high-feature model as
their preferred player. However, a significantly lower per-
centage of the participants who had used the high-feature
model (44%) chose the high-feature model (z = 2.5, p=
.01), even though they had already invested time learning to
use this model. Moreover, participants who used the high-
feature model were less confident in their choices (Mhigh =
4.7) than participants who used the low-feature model
(Mlow = 5.4; F(1, 94) = 5.8, p= .02), and they rated the
choice as more difficult (Mhigh = 3.1) than participants who
used the low-feature model (Mlow = 2.3; F(1, 94) = 5.7, p=
.02). When we controlled for expertise, participants’ confi-
dence in their choices was lower after use (Mafter = 5.0) than
before use (Mbefore = 5.8; F(1, 185) = 14.8, p< .001), sug-
gesting that usage does not enhance confidence in product
evaluations.
Finally, we analyzed the usability data. There was no dif-
ference in the number of tasks completed in the low- and
high-feature conditions (Mlow = 3.2, Mhigh = 3.1; p= .45).
The number of tasks completed was positively correlated
with perceived product usability (r = .30, p< .01). Partici-
pants in the high-feature condition spent marginally more
time completing the four tasks than participants in the low-
feature condition (Mlow = 6.9 minutes, Mhigh = 9.2 minutes;
F(1, 94) = 3.4, p= .07). The amount of time required to
complete the four tasks was negatively correlated with both
participants’ expertise (r = –.31, p< .01) and perceived
product usability (r = –.23, p= .05).
Discussion
The results of Study 3 show that product use structurally
changes consumers’ preferences. In support of our predic-
tions, consumers gave more weight to capability and less
weight to usability in their expected utilities than in their
experienced utilities. After product use, consumers no
longer evaluated the product with the highest number of
features more favorably, supporting the existence of a fea-
ture fatigue effect. Our findings also suggest that con-
sumers’ expertise does not eliminate the feature fatigue

Feature Fatigue 439
9For simplicity, we model this relationship as linear. However, previous
research suggests that the effect of number of features on capability can
have diminishing returns (Nowlis and Simonson 1996). This has no impact
on our results. If we replace the linear formulation in Equation 1 with a
quadratic function, all results are replicated.
effect. The shift in preferences before and after use occurred
just as strongly for experts as for novices. If adding product
features improves the initial attractiveness of a product but
decreases consumers’ satisfaction after using the product,
how should firms address this problem? How many features
should managers offer to consumers?
THE EFFECTS OF PRODUCT FEATURES ON FIRMS’
PROFITS
In this section, we present an analytical model based on
findings from our three studies to provide managerial
insights into the influence of number of features on firm
profitability. We assume a scenario in which adding features
is essentially free (as is the case in many information-based
products), implying that incremental profitability from
adding features equals incremental revenue. Thus, we
express incremental profit (revenue) from number of fea-
tures, R, as a function of number of features, F, for a typical
customer. We decompose R into a positive effect due to
product capability, C, minus a negative effect due to the
lack of usability, D.
From Studies 1–3, we know that perceived product capa-
bility increases with number of features. We also know
from Study 3 that adding features has a smaller effect on
perceived capability after use than before use. For this rea-
son, denoting capability before and after use as C1and C2,
respectively, we model capability as follows:9
(1) C1= dF, and C2= eF
c,d,e > 0 and d > e.
From Studies 1–3, we also know that usability is perceived
to decline with the number of features, and based on Study
1, this decline appears to accelerate. Thus, we model lack of
usability, D, as follows:
(2) D = aF + bF2a, b > 0.
We also impose d,e > a, to ensure that the optimal feature
level will be positive (otherwise, the firm should not create
the product at all).
We now construct a dynamic scenario to help generate
insights into short-term and long-term profitability. For ana-
lytical parsimony and tractability, we adopt a two-period
scenario in which profit from initial purchases is repre-
sented by the first period and profit from all subsequent
purchases are represented by the second period. We weigh
the second period by a factor, w, that represents the weight
with which subsequent purchases contribute to the typical
customer’s net present value. This weight encompasses
such things as the firm’s discount rate, the typical lifetime
of a customer, the length of the firm’s planning horizon, and
the extent to which the product category is conducive to
repeat sales. Denoting profits in the first and second period
as R1and R2, respectively, and the net present value of the
customer’s profit stream as Rtot,wehave, from Equations 1
and 2,
10Results for the effects of competition are available from the authors on
request.
(3) R1= C1– D = (d – a)F – bF2,
R2= C2– D = (e – a)F – bF2, and
Rtot = R1+ wR2= [(d – a) + w(e – a)]F – (1 + w)bF2.
We f irst consider a myopic profit maximization in which
the firm attempts to find the number of features that maxi-
mizes initial choice (and thus initial profits). This amounts
to maximizing R1with respect to F. It is easily shown that
R1is maximized when F1= (d – a)/2b.
Similarly, if the firm attempts to maximize repurchase
(and thus second-period profits), the firm maximizes R2
with respect to F, leading to the optimal value of F2= (e –
a)/2b. If the firm instead attempts to maximize the net pres-
ent value of the customer’s profit stream, as financial ana-
lysts would consider optimal, the firm maximizes Rtot with
respect to F, leading to the optimal value
(4) Fopt = [(d – a) + w(e – a)]/[2b(1 + w)].
If we assume that the firm wants to maximize the net
present value of the typical customer’s profit stream, the
following results arise from these equations:
Result 1: Maximizing initial choice, sales, or profits results in
the inclusion of too many features. It is easily shown
that F1> Fopt > F2, implying that the optimal number
of features is less than the number that maximizes ini-
tial choice, sales, or profits. This casts into doubt the
exclusive reliance on conjoint or discrete choice mod-
els to determine the appropriate product configuration
because the use of such methods, which use initial
choice as their criterion, will typically result in the
inclusion of too many features.
Result 2: Maximizing repurchase results in the inclusion of too
few features. This again results from Fopt > F2.
Result 3: If the number of features is sufficiently large, addi-
tional features should not be added, even if they can
be added at no cost. Specifically, when F > Fopt,
adding additional features reduces the net present
value of the profit stream. Again, this is in contradic-
tion to the typical economic/conjoint utility model,
which suggests that adding features can only add
utility.
Result 4: As the emphasis on future sales increases, the optimal
number of features decreases. This follows from
∂Fopt/∂w= (e – d)/(2b[1 + w]) < 0 because the
denominator is positive and d > e. This result also has
meaning for managers. With business becoming more
relationship focused over time and with increasing
amounts of attention being paid to the lifetime value
of the customer, this result implies that, on average,
firms should make their products simpler to increase
repurchase and future sales.
For simplicity of exposition, the preceding results have
been obtained for a monopolist. However, it can be shown
that the consideration of competition does not change any
of the substantive conclusions.10 The only change is that in
the case of choice inertia (customers choosing a brand tend
to stay with a brand), the optimal number of features
increases. Nevertheless, Results 1–4 still hold.
Figure 3 usefully summarizes the analytical results. As
the figure shows, there is always an optimal number of fea-
tures, but this optimal number depends on the goals of the

440 JOURNAL OF MARKETING RESEARCH, NOVEMBER 2005
Number of Features
Excess
features
Insuffiicient
features
F2Fopt F1
Response Maximize
repurchase
Maximize
net present
value of
customers
Maximize
initial
purchase
Figure 3
NUMBER OF PRODUCT FEATURES AND FIRM OBJECTIVES
firm. If the firm wants to maximize initial choice (or equiv-
alently, profits from initial purchases), the number of fea-
tures F1is optimal. Any greater number of features is
incompatible not only with the initial purchase objective but
also with any profit objective. If the firm wants to maximize
probability of repurchase (or equivalently, profits from
repurchase), F2is the optimal number of features (and any
less is also incompatible with any profit objective). If the
firm wants to maximize the net present value of the cus-
tomer’s profit stream, Fopt is best. The only range of number
of features that the firm should ever consider is the range
from F2to F1, with the optimal level of features dependent
on the nature of the firm’s objective.
GENERAL DISCUSSION
Our goal in this research was to examine the effects of
increasing the number of product features on consumers’
expected and experienced product utilities. In three studies,
we showed that increasing the number of product features
has a positive effect on perceived capability but a negative
effect on perceived usability. Thus, whether adding desir-
able, important features to a product will increase or
decrease utility depends on the relative weights of capabil-
ity and usability in consumers’ utility functions. Study 3’s
results indicate that consumers assign more weight to prod-
uct capabilities in their evaluations before use than after use
and less weight to product usability in their satisfaction rat-
ings before use than after use. Thus, what appears to be
attractive in prospect does not necessarily appear to be good
in practice: When using a product, consumers may become
frustrated or dissatisfied with the number of features they
desired and chose before using the product. In short, prod-
uct capability may become too much of a good thing.
These changes in the relative weights of product capabil-
ity and usability are consistent with our hypotheses based on
construal-level theory. Before using a product, consumers
seem to be more focused on desirability issues, such as the
product’s capabilities, and less focused on feasibility con-
cerns, such as usability, than they are after using a product.
Because different considerations are salient in expected and
experienced utility, using a product can change the structure
of consumers’ preferences. Such changes in preferences are
significant because they suggest that consumers may not
choose products that maximize their long-term satisfaction.
The impact of product usage on preferences is an under-
studied area. Consumer behavior research has traditionally
focused more on prepurchase processes, such as informa-
tion processing and choice (Bazerman 2001). Although the
services literature has long recognized the importance of
relationship duration, ongoing usage levels, and satisfaction
(e.g., Bolton and Lemon 1999), this literature has focused
on changes over time and has not developed theoretical
frameworks to explain why consumers’ underlying prefer-
ences might change.
In some respects, our studies presented a conservative
test of our hypotheses. First, we used college students as
our sample, a demographic segment that tends to be more
open to new technology and new features than other seg-
ments. A recent nationwide survey indicated that after buy-
ing a high-tech product, 56% of consumers are over-
whelmed by its complexity, and this percentage is positively
correlated with age (Rockbridge Associates 2004). Second,
our high-feature product had only 21 features, a relatively
low number of features in some product categories. The
dashboard alone of the BMW 745 has more than 700 fea-
tures. Further research should examine products with more
features and test for nonlinearities in evaluations as features
are added. Finally, our studies considered only features that
added functionality to the product and were reasonably
familiar to the participants. The negative effect of unimpor-
tant or highly complex features may be stronger.
Further research should also examine consumers’ reac-
tions to product features over a longer period of time. For
example, it would be interesting to learn whether con-
sumers attribute poor usability to the large number of fea-
tures in the model they chose or to the brand, potentially
damaging a firm’s sales across multiple categories. Even if
consumers learn about the negative effects of too many fea-
tures after a usage experience, this learning might be forgot-
ten in future purchase situations, when product capability
again becomes the key driver of evaluations. Finally, con-
sumers use various strategies to cope with technology
(Mick and Fournier 1998). If consumers use confrontative
strategies (e.g., mastering, partnering), the effects of prod-
uct features on usability and experienced utility may
decrease over time. However, if consumers use avoidance
strategies (e.g., distancing, abandonment), the effect of
product features on experienced utility is likely to remain
strong.
Although supply-side explanations for the proliferation
of product features abound, our results demonstrate that
demand-side explanations are sufficient for feature fatigue
to occur. It is certainly true that companies often find it
cheaper to produce feature-rich products that can satisfy
the needs of heterogeneous consumers than to produce
more narrowly targeted products with fewer features.
However, companies often add features to products
because they believe that their customers want more fea-
tures. Indeed, our results indicate that even conducting

Feature Fatigue 441
market research may not eliminate the problem. If compa-
nies conduct market research by asking customers to eval-
uate products without using them, too much weight will
be given to capability compared with usability, and it is
likely that too many features will be added to the
products.
MANAGERIAL IMPLICATIONS
Our research has several important managerial implica-
tions. First, our findings call into question the predictive
power of attribute-based models for determining the opti-
mal number of features. Firms planning new products or
considering product improvements typically use market
research techniques such as conjoint analysis or discrete
choice analysis. The conjoint model, for example, defines
the product as a bundle of attributes and estimates part-
worths for each attribute. Because market shares are pre-
dicted on the basis of these partworths, each positively val-
ued feature increases a product’s market share compared
with products without the feature. Our results suggest that
traditional conjoint analysis can lead to marketing myopia,
in which firms maximize initial sales. This occurs because
usability, a global rather than an attribute-based characteris-
tic, is underweighted by consumers before product use but
becomes a critical element in consumers’ satisfaction dur-
ing use. Our results indicate that a product use experience
may be required to increase the salience of usability so that
its relevance in choice approaches its relevance in use.
Thus, consumers’ preferences may be more accurately pre-
dicted using customer-ready prototypes and product-in-use
research (Srinivasan, Lovejoy, and Beach 1997).
Because additional features can differentiate a product
from competitors (Carpenter, Glazer, and Nakamoto 1994)
and add desired functionality, the benefits of adding new
features to products are evident. However, managers rarely
consider the full cost of adding features. The financial costs
of adding new features are typically weighted more heavily
than intangible customer usability costs. Thus, as the mar-
ginal cost of adding features decreases, approaching zero
for information-based products (e.g., software), firms are
likely to increase product capability beyond the optimal
level. This is a dangerous trend: Both our empirical findings
and our analytical model suggest that adding costless fea-
tures can damage firms’ profitability by decreasing the
usability of products and consumers’ satisfaction with them.
As firms shift their focus from one-time transactions to
long-term relationships with customers, the importance of
product usability—and the research methods that enable
managers to incorporate usability into their decision mak-
ing—will only increase. Our results suggest that too many
features can encourage initial purchase but damage satisfac-
tion and reduce repurchase probabilities, leading to lower
customer lifetime values. As our model demonstrates, the
optimal number of features varies depending on firms’
objectives. As firms’ reliance on continuing customer rela-
tionships increases, the optimal number of features
decreases, implying that products should be made simpler.
Conducting research with customers both before and after
they use products will enable firms to measure product-
specific parameters for the model and, thus, to predict the
number of features that will maximize their long-term
profits.
What can firms do to minimize feature fatigue? Our
findings suggest that managers should consider offering a
wider assortment of simpler products rather than all-
purpose, feature-rich products. Instead of packing one
model with many features to address market heterogene-
ity, firms might enhance consumer satisfaction by devel-
oping more tailored products with limited sets of capabili-
ties that appeal to different segments. Consumers can now
purchase a single product that functions as a cell phone,
game console, calculator, text-messaging device, wireless
Internet connection, personal digital assistant, digital cam-
era, MP3 player, and global positioning system. However,
although purchasing this highly complex product may
give the consumer bragging rights, each function the con-
sumer does not actually use adds to the difficulty of learn-
ing to use the product without providing any functional
benefit.
A challenge of creating and marketing more narrowly
targeted products is that choosing among a wider variety of
products can be more difficult for consumers (Schwartz
2004). Rather than using the heuristic of buying features
they may need (but are not sure they will need), consumers
will need to consider carefully which features to purchase.
Moreover, our empirical results suggest that during the
choice process, consumers are tempted by products that
offer greater capability. To minimize feature fatigue, deci-
sion aids, such as recommendation agents that help con-
sumers choose the right products for their needs, could be
designed to increase the salience of usability. Offering
extended product trials also may help consumers learn
which products best suit their needs by increasing the
salience of product usability. For example, the companies
that sell digital media players RealOne and WinAmp offer
evaluation versions of their products. By decreasing the gap
between consumers’ preferences during choice and use,
such strategies may increase both customer satisfaction and
customer lifetime value.
REFERENCES
Alba, Joseph W. and J. Wesley Hutchinson (1987), “Dimensions of
Consumer Expertise,” Journal of Consumer Research, 13
(March), 411–38.
Ammirati, Sean (2003), “Other Voices: Ask Your Users: Less
Really Can Be More,” Information Week, (July 14), (accessed
July 17, 2003), [available at http://www.informationweek.com/
story/showArticle.jhtml?articleID=12800101].
Bazerman, Max H. (2001), “Consumer Research for Consumers,”
Journal of Consumer Research, 27 (March), 499–504.
Bolton, Ruth N. and Katherine N. Lemon (1999), “A Dynamic
Model of Customers’ Usage of Services: Usage as an
Antecedent and Consequence of Satisfaction,” Journal of Mar-
keting Research, 36 (May), 171–86.
Brown, Christina L. and Gregory S. Carpenter (2000), “Why Is the
Trivial Important? A Reasons-Based Account for the Effects of
Trivial Attributes on Choice,” Journal of Consumer Research,
26 (March), 372–85.
Carpenter, Gregory S., Rashi Glazer, and Kent Nakamoto (1994),
“Meaningful Brands from Meaningless Differentiation: The
Dependence on Irrelevant Attributes,” Journal of Marketing
Research, 31 (August), 339–50.
Celsi, Richard L. and Jerry C. Olson (1988), “The Role of Involve-
ment in Attention and Comprehension Processes,” Journal of
Consumer Research, 15 (September), 210–24.

442 JOURNAL OF MARKETING RESEARCH, NOVEMBER 2005
Chin, John P., Virginia A. Diehl, and Kent L. Norman (1988),
“Development of an Instrument Measuring User Satisfaction of
the Human-Computer Interface,” in Human Factors in Comput-
ing Systems, CHI’88 Conference Proceedings. New York: Asso-
ciation for Computing Machinery, 213–18.
Freund, Bruno, Herbert König, and Norbert Roth (1997), “Impact
of Information Technologies on Manufacturing,” International
Journal of Technology Management, 13 (3), 215–28.
Goldenberg, Jacob, Roni Horowitz, Amnon Levav, and David
Mazursky (2003), “Finding Your Innovation Sweet Spot,” Har-
vard Business Review, 81 (March), 120–29.
Jones, Russell A. (1977), Self-Fulfilling Prophecies: Social, Psy-
chological, and Physiological Effects of Expectations. Hillsdale,
NJ: Lawrence Erlbaum Associates.
Lancaster, Kelvin (1971), Consumer Demand: A New Approach.
New York: Columbia University Press.
Liberman, Nira and Yaacov Trope (1998), “The Role of Feasibility
and Desirability Considerations in Near and Distant Future
Decisions: A Test of Temporal Construal Theory,” Journal of
Per sonality and Social Psychology, 75 (1), 5–18.
McLaughlin, Janice and David Skinner (2000), “Developing
Usability and Utility: A Comparative Study of the Users of New
IT,” Te chnology Analysis & Strategic Management, 12 (3),
413–23.
Mick, David G. and Susan Fournier (1998), “Paradoxes of Tech-
nology: Consumer Cognizance, Emotions, and Coping Strate-
gies,” Journal of Consumer Research, 25 (September), 123–43.
Mitchell, Andrew A. and Peter A. Dacin (1996), “The Assessment
of Alternative Measures of Consumer Expertise,” Journal of
Consumer Research, 23 (December), 219–39.
Mukherjee, Ashesh and Wayne D. Hoyer (2001), “The Effect of
Novel Attributes on Product Evaluation,” Journal of Consumer
Research, 28 (December), 462–72.
Nielsen, Jacob (1993), Usability Engineering. San Diego: Acade-
mic Press.
Nowlis, Stephen M. and Itamar Simonson (1996), “The Effect of
New Product Features on Brand Choice,” Journal of Marketing
Research, 33 (February), 36–46.
Olson, Jerry C. and Thomas J. Reynolds (1983), “Understanding
Consumers’ Cognitive Structures: Implications for Advertising
Strategy,” in Advertising and Consumer Psychology, L. Percy
and A.G. Woodside, eds. Lexington, MA: Lexington Books,
77–90.
Peracchio, Laura A. and Alice M. Tybout (1996), “The Moderating
Role of Prior Knowledge in Schema-Based Product Evalua-
tions,” Journal of Consumer Research, 23 (December), 177–91.
Rockbridge Associates (2004), National Technology Readiness
Survey. Falls Church, VA: Rockbridge Associates.
Schwartz, Barry (2004), The Paradox of Choice: Why More Is
Less. New York: HarperCollins.
Simonson, Itamar, Ziv Carmon, and Suzanne O’Curry (1994),
“Experimental Evidence on the Negative Effect of Product Fea-
tures and Sales Promotions on Brand Choice,” Marketing Sci-
ence, 13 (Winter), 23–40.
Spence, Mark T. and Merrie Brucks (1997), “The Moderating
Effects of Problem Characteristics on Experts’ and Novices’
Judgments,” Journal of Marketing Research, 34 (May), 233–47.
Srinivasan, V., William S. Lovejoy, and David Beach (1997),
“Integrated Product Design for Marketability and Manufactur-
ing,” Journal of Marketing Research, 34 (February), 154–63.
Wiklund, Michael (1994), Usability in Practice: How Companies
Develop User-Friendly Products. San Diego: Academic Press.
Ziefle, Martina (2002), “The Influence of User Expertise and
Phone Complexity on Performance, Ease of Use and Learnabil-
ity of Different Mobile Phones,” Behaviour & Information
Technology, 21 (5), 303–311.