Content uploaded by Fabian Reinkemeier
All content in this area was uploaded by Fabian Reinkemeier on Mar 09, 2021
Content may be subject to copyright.
VOICE APPS IN VOICE COMMERCE: THE IMPACT OF HUMANNESS ON
PARASOCIAL INTERACTION AND RELATIONSHIP QUALITY
Fabian Reinkemeier, University of Goettingen
Waldemar Toporowski, University of Goettingen
Contact Information: For further information, please contact Fabian Reinkemeier,
University of Goettingen (Fabian.Reinkemeier@wiwi.uni-goettingen.de).
Keywords: voice assistant, voice commerce, social response theory, social cue, humanlike
Description: This research examines the impact of more humanlike design in voice apps on
parasocial interaction and relationship quality in voice commerce.
Recent years have seen rapid growth in the popularity of voice assistants, enabling companies
to develop their own voice apps and establish voice commerce as an additional sales channel.
Within such apps, firms can create a virtual voice agent (VVA) that allows them to enter into
an online sales dialogue with customers using natural spoken language. Despite its manifold
advantages, voice commerce still faces the familiar challenges that afflict online shopping
and technology adoption: Customers are generally uncomfortable with the interaction and
have difficulties establishing satisfying relationships with voice apps, leading them to use
voice commerce only sporadically. This study explores how these obstacles could be
A machine possessing humanlike design features, referred to as social cues, can foster more
humanlike and social interactions with users (Nass and Moon 2000; Nass, Steuer, and Tauber
1994). Accordingly, implementing such features in VVAs might enable them to be socially
present in interactions and build social relationships with customers in voice commerce.
Therefore, we pose the following research question:
RQ: To what extent do more humanlike design features in VVAs promote parasocial
interaction and thus enhance the customer VVA relationship in voice commerce?
Method and Data
To answer this question, we conducted a laboratory experiment with 323 participants,
applying a between-subjects design with two versions of a VVA. These versions were
identical (e.g., in terms of dialogue flow) except that the treatment condition included a
combined set of verbal, auditory, and invisible social cues. To carry out the experiment, we
. Participants were randomly
assigned to one of the two conditions and were tasked to order a specific book in real-time
via VVA. By manipulating the independent variable, coding a high level of humanness as 1
and a low level as 0, we also successfully ascertained whether the differences in human
perception are valid. We tested the research model using PLS-SEM via SmartPLS 3.2.9
software and a bootstrap resampling method with 5,000 samples (Ringle, Wende, and Becker
Summary of Findings
To determine what might help overcome barriers to voice commerce, we investigated the
effects of humanizing a voice app by implementing various humanlike design features. The
data suggests that a stronger integration of such features
social interaction and quality of relationship with the voice app. Nevertheless, our analysis
indicates that humanizing voice apps can also elaborate negative effects. Overall, rendering
the human machine dialogue more natural could still be the key to a broad acceptance of
Our study offers a threefold contribution: First, we empirically extend knowledge on the
impact of humanness in interaction with voice assistants. Various scholars have called for
such research (e.g., Gnewuch et al. 2018; Seeger, Pfeiffer, and Heinzl 2018), as until recently,
a conversation between human and machine with voice as the input and output was
impossible. In this regard, we demonstrate that including more social cues leads to greater
perceptions of humanness in a VVA. Achieving this state has an overall positive impact on
interaction; customers tend to regard such voice apps as more socially present. Second, our
study contributes to literature on relationship marketing by demonstrating that perceiving
social presence and parasocial interaction in voice commerce improves relationship quality.
Third, our results offer important implications for companies and their design of voice apps.
We suggest that voice app designers should implement various social cues with the goal of
rendering VVAs more humanlike. Despite our analysis revealing that more humanness can
also induce certain negative effects, the extensive use of humanlike design features can have
overall and spark a budding
relationship with the voice app.
References are available on request.