A State-of-The-Art Overview and Future Research Avenues

for Self-Service Business Intelligence and Analytics

ECIS 2020
Sven Michalczyk, Mario Nadj, Darius Azarfar, Alexander Mädche, Christoph Gröger
INSTITUTE OF INFORMATION SYSTEMS AND MARKETING (IISM)
RESEARCH GROUP INFORMATION SYSTEMS AND SERVICE DESIGN (ISSD)

KIT – The Research University in the Helmholtz Association

www.kit.edu
 Business Intelligence & Analytics (SSBIA)

…encourages users to prepare, analyze and consume data on their own, with nominal IT
support (based on Gartner IT Glossary, 2019; Lennerholt et al., 2018)
“empower casual users to perform analytics” and “power users can accomplish their task
[…] more easily and quickly than before”. (Alpar & Schulz 2016)
Addresses the challenge to support different user roles with diverse levels of expertise
for various analytical demands ranging from (1) information usage, (2) over information
creation, (3) towards information resource creation (Alpar & Schulz 2016).

2 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Motivation
Size of the Global BIA Market (BARC, 2019)
▪ Adoption of SSBIA is challenging due to technical 35 29,84
or organizational boundaries (Logi 2017) 30

Billionen US. $
25
▪ Vast number of studies requires structuration 20 15,64
15
▪ Lack of a systematic approach to illustrate the 10
state of-the-art along the various characteristics of 5
0
SSBIA
2016 2022
▪ Although some scholars sketched out SSBIA
Number of SSBIA
challenges (e.g., Lennerholt et al., 2018) future Related Scientific Publications
25
reserach needs more guidance to be well-directed 20
20

15

Count
What are the potential avenues in 10 8
6 7 6
SSBIA for future research? 5
1 1 2 2 1
3 4

0
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019

3 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Systematic Literature Review
Following Webster and Watson (2002), Kitchenham and Charters (2007)

Search String Hit Statistics

1 “Self*Service” AND ( Step 2.1: Search (# 809) WebOfScience (# 203)

Scopus (# 472)
IEEE (# 132)
2 “Business Intelligence” OR “Analytic*” arXiv & SSRN (# 2)

3 OR “Information System*” OR “Decision Support System*” Study title & abstract (#58)
Step 2.2: Select (#60)
4 OR “Machine Learning” OR Study full text (#49)

5 “Data” AND (“Understanding” OR “Cleaning” OR Step 2.3: Extract / Analyze Forward/backward (#60)

“Preparation” OR “Integration” OR “Wrangling” OR

“Modeling” OR “Mining” OR
“Discovery” OR “Exploration” OR “Visualization”)

4 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Conceptual Framework
De Mauro et al., 2018;
1 Lismont et al., 2019 1 Alpar & Schulz, 2016 2
Classification Process
Self-Service
1.
1 Derived deductivley from User Role Experience Perspective
Levels
existing literature
Usage of
2.
2 Cross-validate existing Casual User Novice Information Artifact-Centric
Ressources
codes or create new ones
Creation of
3.
3 Assign codes Power User Expert
Information
User-Centric

independently by two
researchers and resolve Creation of
Governance-
IT-User 2 Information
Centric
conflicts Resources
3

5 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 User Roles
Power User
77%
68%

37%
18%

Casual User Data Analyst Data Scientist IT User

Business User & Domain Experts Responsible for transforming Less business and domain Engineers & Developers
data into actionable knowledge (Eckerson, 2012)
Solely consume data Providing necessary data
insights for casual users
(monitoring of dashboards, Structured, unstructured, infrastructure (e.g. data
(Eckerson, 2012)
preparing reports) internal and external data streams via pipelines), thus
Distinguished by data (Abelló et al., 2013) act as enabler (De Mauro et
Scares analytical but extensive
production (clean and al., 2018, Zehnder & Riemer,
business or engineering Applying advanced analytics
aggregate data, De Mauro et 2018)
knowledge (Spahn et al., 2008) (identifying patterns, applying
al., 2018) and data
context and intelligence; De Mgt. and governance of data
Naturally develop a data-driven consumption tasks (analyze
Mauro et al., 2018) (Bilalli et al., 2016)
mindset over time supported by ML outputs; Dinsmore, 2016)
a ontology to learn BIA-related Implementation of data-
Unlike DS, often bound to a
knowledge (Englmeier & Román, intensive systems (Eckerson,
department (Eckerson,
2014) 2012)
2012)

6 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Perspectives 82%

45%
25%

Artefact-centric View User-centric View Governance-centric

Continuous interaction without any Aim to make casual user more Data Lakes (typically not fully IT-
mediation or intervention (Abelló et al. knowledgable (Imhoff & White, 2011) curated) become an important direct
2013), source for SSBIA (Llave,2018; Vo et. al.,
Implementation of information levels
2018),
I.e. Fusion Cubes or (data consumptions tasks, data
production tasks, user profiles, Data quality needs to be sufficient for
Semantic query designer tool (Spahn
availability; Alpar & Schulz, 2016) the analytical task (Stodder, 2015).
et al. 2008)
Crowd-sourced annotation systems
Multi-tool, multi-user platform to create
Leverage ML by a domain-specific may rate the value of reports and data
synergies between casual and power
language (Zehnder and Riemer 2018) sources (Imhoff & White, 2011).
user (Convertino & Echenique)
Too strict governance leads to shadow
IT infrastructure built up by business
departments (Stodder, 2015).

7 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Self-Service Levels
(1) Usage of Information (2) Creation of (3) Creation of Information Ressources
Resources Information
77%
67%
52% 57%
47%
28%
18%

Access to Reports Access to Drill Creation of Access to Creation of Harnessing of Deployment of

"Anywhere" Reports Analytical Mashups New Data Analytical
Reporting Functions Sources Functions
Acces to reports requires the lowest Data access and availability are a Creation of Mashups refers to various data
level of self-reliance and system key concern for information creation preparation tasks. SSBIA should give users the
support from casual users (Alpar & (Berthold et al., 2017). autonomy to explore and integrate new data in
Schulz, 2016). structured and unstructured formats (Bani-Hani et
Because power user are scares in the
al., 2018).
Sharing and collbaboration are market, causal user should be
important system features enabled through SSBIA to fit models Learned statistical models can only create value if
(Marjanovic, 2015) (Dinsmore, 2016). Model they are integrated into business processes. For
management warehouses could e.g. , more technical roles may deploy analytical
Technical solutions to drill-down and
ensure reuse of pre trained models functions to enable casual users to leverage ML
roll-up anywhere functionalities
(Chard et al., 2019). (Zehnder and Riemer, 2018).
(Chouder et al., 2017).

8 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Future Research Avenues

1 Understand the Trade-Off between Top-Down and Bottom-Up SSBIA Capabilities

2 Define a Method for Implementing SSBIA within Organizations

3 Investigate Success Factors for Data Management and Data Governance

4 Support Casual Users‘ Self-Reliance within the Analytical Investigation Process

5 Develop Effective Multi Sensory User Interfaces for Immersive Collaboration Environments

9 Sven Michalczyk Institute of Information Systems and Marketing (IISM)

 Vielen Dank für Ihre Aufmerksamkeit!

Many thanks for your attention!

10 Sven Michalczyk Institute of Information Systems and Marketing (IISM)