Unstructured data in marketing Balducci et al (2018)
Dit artikel gaat over unstructured data (80% van de data bij een bedrijf gemiddeld). Dit is al de data
waar men geen bestemming voor geeft of lastiger te analyseren is door gebruik van een
datamodel. Bijvoorbeeld gesprekken, gebarentaal of klantgedachten zijn moeilijk te analyseren.
Echter is deze data zeker de moeite waard om te bekijken omdat het de besluitvorming in een
bedrijf kan verbeteren en de klantgedragingen beter kan volgen.
The rise of unstructured data (UD), propelled by novel technologies, is reshaping markets and the
management of marketing activities. Yet these increased data remain mostly untapped by many
firms, suggesting the potential for further research developments. The integrative framework
proposed in this study addresses the nature of UD and pursues theoretical richness and
computational advancements by integrating insights from other disciplines. This article makes three
main contributions to the literature:
1) Offering a unifying definition and conceptualization of UD in marketing
2) Bridging disjoint literature with an organizing framework that synthesizes various subsets of
UD relevant for marketing management through an integrative review
3) Identifying substantive, computational, and theoretical gaps in extant literature and ways
to leverage interdisciplinary knowledge to advance marketing research by applying UD
analyses to underdeveloped areas.
Wikipedia definition: UD is commonly understood as Binformation that either does not have a
predefined data model or is not organized in a pre-defined manner.
An estimated 80% of data held by firms today are unstructured (Rizkallah 2017), and they are
growing 15 times faster than structured data
Unlocking the insights embedded in this burgeoning resource has the potential to be
particularly valuable in marketing, sales and service settings where UD volumes are an
estimated five times greater than SD
Definition of UD in this article: Single data unit in which the information offers a relatively concurrent
representation of its multifaceted nature without predefined organization or numeric values.
Example: Retailers may collect many data points about a customer during a purchase transaction, but
such information comes from multiple data units (e.g., time, cost, location of purchase), so each unit
must be separately considered and prepared for quantitative analysis.
, Characteristics of UD:
1) UD is nonnumeric. They lack predefined numeric assignments for the constructs of interest
and researchers must conduct manual or automatic coding prior to analysis.
Example: To determine the level of customer expressed affect through nonverbal cues in a service
exchange, a researcher must first consider which nonverbal cues embody different levels of positive,
negative, and neutral affect before determining the degree of expressed affect in each cue and
counting the number of occurrences in the interaction.
2) UD is multifaceted. A single unit of highly UD possesses multiple facets, each offering unique
information enabling the researcher to select and analyze facet(s) based on the research
goals.
Example: Voice data contains many facets (e.g., pitch, speech rate, intensity) that all provide unique
information since each of these facets conveys different information about the speaker (e.g.,
affective state, persuasiveness).
3) UD maintains concurrent representation. The simultaneous presence of a single data unit’s
multiple facets that each provide unique information allows an UD unit to represent different
phenomena at the same time. Thus, the scholar can examine diverse research questions with
a single highly UD unit through examination of the concurrent flow of these unique facets.
Example: Consider text data documenting an email exchange between a salesperson and a customer.
A single unit of this text data contains many unique facets (e.g., syntax, semantics) that occur
simultaneously, and each of these facets provides distinct information that the scholar can use to
assess different phenomena (e.g., persuasion, affect)
Leveraging unstructured data for unique theoretical insights: We present three general categories
of theoretical contribution that can be derived from UD to create, communicate, and deliver value to
customers.
Dit artikel gaat over unstructured data (80% van de data bij een bedrijf gemiddeld). Dit is al de data
waar men geen bestemming voor geeft of lastiger te analyseren is door gebruik van een
datamodel. Bijvoorbeeld gesprekken, gebarentaal of klantgedachten zijn moeilijk te analyseren.
Echter is deze data zeker de moeite waard om te bekijken omdat het de besluitvorming in een
bedrijf kan verbeteren en de klantgedragingen beter kan volgen.
The rise of unstructured data (UD), propelled by novel technologies, is reshaping markets and the
management of marketing activities. Yet these increased data remain mostly untapped by many
firms, suggesting the potential for further research developments. The integrative framework
proposed in this study addresses the nature of UD and pursues theoretical richness and
computational advancements by integrating insights from other disciplines. This article makes three
main contributions to the literature:
1) Offering a unifying definition and conceptualization of UD in marketing
2) Bridging disjoint literature with an organizing framework that synthesizes various subsets of
UD relevant for marketing management through an integrative review
3) Identifying substantive, computational, and theoretical gaps in extant literature and ways
to leverage interdisciplinary knowledge to advance marketing research by applying UD
analyses to underdeveloped areas.
Wikipedia definition: UD is commonly understood as Binformation that either does not have a
predefined data model or is not organized in a pre-defined manner.
An estimated 80% of data held by firms today are unstructured (Rizkallah 2017), and they are
growing 15 times faster than structured data
Unlocking the insights embedded in this burgeoning resource has the potential to be
particularly valuable in marketing, sales and service settings where UD volumes are an
estimated five times greater than SD
Definition of UD in this article: Single data unit in which the information offers a relatively concurrent
representation of its multifaceted nature without predefined organization or numeric values.
Example: Retailers may collect many data points about a customer during a purchase transaction, but
such information comes from multiple data units (e.g., time, cost, location of purchase), so each unit
must be separately considered and prepared for quantitative analysis.
, Characteristics of UD:
1) UD is nonnumeric. They lack predefined numeric assignments for the constructs of interest
and researchers must conduct manual or automatic coding prior to analysis.
Example: To determine the level of customer expressed affect through nonverbal cues in a service
exchange, a researcher must first consider which nonverbal cues embody different levels of positive,
negative, and neutral affect before determining the degree of expressed affect in each cue and
counting the number of occurrences in the interaction.
2) UD is multifaceted. A single unit of highly UD possesses multiple facets, each offering unique
information enabling the researcher to select and analyze facet(s) based on the research
goals.
Example: Voice data contains many facets (e.g., pitch, speech rate, intensity) that all provide unique
information since each of these facets conveys different information about the speaker (e.g.,
affective state, persuasiveness).
3) UD maintains concurrent representation. The simultaneous presence of a single data unit’s
multiple facets that each provide unique information allows an UD unit to represent different
phenomena at the same time. Thus, the scholar can examine diverse research questions with
a single highly UD unit through examination of the concurrent flow of these unique facets.
Example: Consider text data documenting an email exchange between a salesperson and a customer.
A single unit of this text data contains many unique facets (e.g., syntax, semantics) that occur
simultaneously, and each of these facets provides distinct information that the scholar can use to
assess different phenomena (e.g., persuasion, affect)
Leveraging unstructured data for unique theoretical insights: We present three general categories
of theoretical contribution that can be derived from UD to create, communicate, and deliver value to
customers.
