Gr Interakt Org (2023) 54:65–77

https://doi.org/10.1007/s11612-023-00666-9

HAUPTBEITRÄGE - THEMENTEIL

Data-driven information for action

Kristin Wulff1,2 · Hanne Finnestrand1

Accepted: 19 December 2022 / Published online: 21 February 2023

© The Author(s) 2023

Abstract
Because of the increase in data and the possibilities created by machine learning, organizations are now looking to become
more data-driven. In sociotechnical systems design there has been a focus on designing information for action to support
decentralized organizations. The purpose of this article, published in Gruppe. Interaktion. Organisation. is to discuss how
data may be gathered and used in organizations striving to become data-driven.
Explorations are based on interviews with experts (leaders and designers) in 13 organizations working on becoming more
data-driven.
This study points to 4 findings: first, if someone is expected to record data that informs other people’s actions can lead
to data quality issues, which can be mitigated by providing transparency or supporting a joint information for action as
an organizational design choice. Second, as organizations are becoming more data-driven, many tasks performed in the
organization become design-related. This influences the type of data recorded and used for action. Third, more of the
people in the organizations engage in designing the information for action for themselves and others, which means that
they might need reskilling. Fourth, the boundaries of what can be considered information for action and for whom should
by explored and reflected upon by the people involved in the (re)design.
This means that, as organizations strive to become data-driven, the sociotechnical principle of information flow becomes
a central challenge. To ensure quality organizations, there is a need to upskill or reskill employees so that they are able to
design and use data for action.

Keywords Data-driven · Machine learning · Sociotechnical systems design · Organizational design · Information for
action

Datengesteuerte Information zum Handeln

Zusammenfassung
Aufgrund der Zunahme von Daten und der Möglichkeiten, die Maschinelles Lernen schafft, versuchen Organisationen
nun mehr und mehr datengesteuert zu werden. Um dezentralisierte Organisation(en) zu unterstützen, hat soziotechnische
Systemgestaltung den Schwerpunkt auf die Gestaltung von Information zum Handeln gelegt. Der Zweck dieses Artikels,
veröffentlicht in Gruppe. Interaktion. Organisation., ist zu erörtern, wie Organisationen, die bestrebt sind, datengesteuert
zu werden, Daten sammeln und verwenden können.
Die Untersuchungen basieren auf Interviews mit Experten (Führungskräften und Designern) in 13 Organisationen, die an
vermehrter Datensteuerung arbeiten.

 Kristin Wulff
[email protected]

1
Faculty of Economics and Management, Department
of Industrial Economics and Technology Management,
Norwegian University of Science and Technology,
2
Trondheim, Norway Kantega AS, Trondheim, Norway

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
66 K. Wulff, H. Finnestrand

Die Studie weist auf vier Erkenntnisse hin: Erstens, wenn von jemandem erwartet wird, Daten zu erfassen, die das
Handeln von Anderen beeinflussen, kann dies zu Datenqualitätsproblemen führen, die durch eine Gestaltungswahl von
Transparenz und der Unterstützung von gemeinsamer Information zum Handeln abgemildert werden können. Zweitens, im
Laufe vermehrter Datensteuerung werden die in Organisationen ausgeführten Aufgaben zunehmend gestaltungsbezogen.
Dies beeinflusst die Art der erfassten und zum Handeln verwendeten Daten. Drittens, da mehr Beschäftigte Information
zum Handeln für sich selbst und andere gestalten, müssen diese möglicherweise weitergebildet werden. Viertens, die
Grenzen dessen, was als Information zum Handeln angesehen wird und für wen, sollte von den Beschäftigten die an der
(Neu-)Gestaltung beteiligt sind sondiert und überdacht werden.
Das bedeutet, wenn Organisationen danach streben, datengesteuert zu werden, wird der soziotechnische Ansatz zum
Informationsfluss zur zentralen Herausforderung. Um die Qualität in und von Organisationen zu gewährleisten, müssen
Mitarbeiter weitergebildet oder umgeschult werden, um Daten zum Handeln gestalten und nutzen zu können.

Schlüsselwörter Datengesteurt · Maschinelles Lernen · Soziotechnische Systemgestaltung · Organisationsdesign ·

Information zum Handeln

1 Introduction and theoretical positioning by Govers and van Amelsvoort (2018) where processes are
parallelized and supported with one IT information inter-
Customer expectation around fast delivery and personaliza- face each, in what they call Archipelago architecture. One
tion of services is increasing. Insurance payouts are often important design principle advocated by Cherns (1987), and
received seconds after entering the insurance claim, and we reinforced by others, is access to information for action. For
expect to receive fast feedback when applying for a car loan. the people in the organization who need it in order to carry
The possibilities to personalize and automatize are based on out their work and to handle the variance they experience
the increase in stored data because of the digitization of data it is crucial that the information reaches them.
and digitalization of work (Gobble 2018). Such data, paired Organizations have always used data to help them make
with the possibilities of Machine Learning (ML), has cre- decision, but what is new is the increased volume and vari-
ated all kinds of iconic personalized services, from the Net- ety of data, the possibility to share the data, and its use in
flix Recommendation Engine to picture recognition systems near real-time (Bean 2021). To be able to use the data in
that can read the label of wine bottles in photos. Accord- innovative ways, organizational learning is crucial (Mikalef
ing to Brynjolfsson and McAfee (2014), organizations are and Krogstie 2020). The data is explored to find ways of
facing the largest transformation since the industrial rev- increasing the competitive advantage of an organization
olution. Digital transformation will alter expertise, change (Provost and Fawcett 2013). Till now, the use have been
work design and boundaries, and enable new ways to co- mainly for the existing business (Brock and von Wangen-
ordinate and control (Faraj et al. 2018). An essential part heim 2019) to improve customer experiences, create new
of the digital transformation is to use data for predictions products and services, make better decisions, make the
and to prescribe action (Iansiti and Lakhani 2020). This can work processes more efficient, and reduce costs (Benbya
be referred to as data-driven decision-making (Brynjolfsson et al. 2020). Data have already inspired new business mod-
and McElheran 2016), and/or becoming a data-driven or- els, and more is expected to in the years to come (Teece
ganization (Berndtsson et al. 2018). When redesigning such and Linden 2017). The improvements in decision-making
organizations it is vital to use design guidelines that result because of an increased use of data can be seen on strate-
in quality work and a flexible organization. gic, tactical and operational decisions (Halper and Stodder
When digital technology is used to change the sociotech- 2017).
nical structures, it can be defined as digitalization. A digital The problems in creating a data-driven organization are
transformation happens when the digitalization and/or dig- mainly non-technical, for instance whether to use data to
ital innovation over time transforms how business is con- take decisive action or whether data is trusted (Halper and
ducted and this transformation changes an entire organiza- Stodder 2017). The data quality is also often a challenge
tion or industry (Osmundsen et al. 2018). To provide guide- (Janssen et al. 2017). To improve the organizational use of
lines for digital transformations, sociotechnical systems de- data, the suggestion is to increase data literacy in the or-
sign (STSD) with its design principles and parameters can ganization (Berndtsson and Svahn 2020). However, there
be used (Babüroğlu and Selsky 2021; Pasmore et al. 2019) is also a need to be able to use the data to improve the
to improve the quality of work (de Sitter et al. 1997; Van Ei- way of working. This can be achieved by testing alterna-
jnatten 1993; Vriens and Achterbergh 2011). STSD can tive response patterns and using data regarding outcome,
also inform the IT architecture, as seen in the description

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Data-driven information for action 67

thus selecting the most appropriate way of working (Herbst between information flow and variance. Information for ac-
1974). tion is used by, for instance, Knight and Parker (2021) who
Unfortunately, the fast and potentially very profitable refer to it with regards to the need for timely feedback to
technological innovations may tempt organizations to take the persons doing the job, and Meacham (2022) who uses it
shortcuts (Zuboff 2019) and/or to choose to have a less con- when researching fire safety. Information for control can be
scious relationship with how a data-driven organization can seen in what is called Algorithmic Management, where al-
help build quality of work. Building a good quality work gorithms, for instance, monitor and evaluate processes, for
system is therefore a design choice. Sociotechnical systems instance the performance of workers. A mediating effect
design (STSD) theory has been reintroduced as a possi- when using Algorithmic management, inspired by STSD,
ble way to understand and design the digital transformation is transparency, perceived fairness and the worker’s possi-
(Babüroğlu and Selsky 2021; Pasmore et al. 2019). Makar- bility to control the system (Parent-Rocheleau and Parker
ius et al. (2020) describes how STS can aid in designing 2022).
the different roles that humans can fill vis-à-vis AI, for in- Handling the variance of the market is seen by many as
stance, controller, collaborator, or conductor. Claussen et al. a managerial task (Emery and Thorsrud 1976), and with the
(2019) suggests that the digitalization calls for “STS-be- increase in data comes the possibility to use insights deliv-
yond” where the paradox of power-related control versus ered by business analytics to support managerial decision-
trust-based control is included as a deliberation in the de- making (Delen and Ram 2018). Following this perspective,
sign of technology. Haga (2019) illustrates how the deliber- the managers in an organization would expect the informa-
ations and dialogue described by Pava are a beneficial way tion for action to be handed to them. This view contradicts
for the organization to create participative processes when the design parameters and principles of STSD, which gives
designing and redesigning technology (Pava 1983, 1986). the information for action to the people doing the work.
Central to STSD is the work unit’s ability and autonomy to Designing the control system bottom-up is a vital part of
deal with the variance it experiences. This is necessary to the STSD, for instance, the STSD by de Sitter et al. (1997).
ensure quality of work and an increased innovation poten- The idea that management make all the decisions in the
tial (de Sitter et al. 1997; Vriens and Achterbergh 2011). organization is truly challenged when data becomes widely
From work in the coal mines (Trist 1981; Trist and Bam- available and this leads to a shift in power from manage-
forth 1951) to software development (Nerur et al. 2010), ment to for instance, analytics experts (Galbraith 2014). By
the variance caused by unpredictability of the context and combining the data into data products that can be made
content impacts how the work of an organization should available to all, the possibility to share real-time data both
be designed (Boonstra and Reezigt 2019). STS designs are internally and externally increases (Vidgen et al. 2017). This
aimed at improving the quality of work and organization means that information that managers may have consid-
through adaption of contents and integration of technology ered to be theirs, may now be given to other people in the
and human tasks (Van Eijnatten 1993). In order to succeed, organization, and also to people outside the organization,
humans need access to information (Cherns 1987). for instance, startups. As Mitki et al. (2019) demonstrated,
To enable the people in the organization to make deci- this sharing of “managerial” information to workers can be
sions and take action based on data, the data must be put into very beneficial for the development of the company. The
a relevant context (Davenport et al. 2012). It must become access to information is central in turbulent environments
information and reach the people that can use it, in a decen- (Govers and Südmeier 2016), and by providing such in-
tralized organizational design (Brödner and Latniak 2003). formation one can develop a learning organization (Herbst
Cherns (1976) states that the information should be de- 1974, 1993). That is, an organization that is able to use data
signed so that the information flows to the people who need to perform double-loop learning (Argyris and Schon 1974)
it to do their job, and not to managers or others who might where the people doing the work is informed in a way that
use it for control. In a later article, Cherns (1987) specifies gives them the opportunity to try new ways responding.
more clearly the three uses of information. Information can Still, it may be difficult for managers to relinquish control
be used for control, it can be used for record, and it can and give the decision-making responsibility based on data
be used for action. Control is about using information to to workers. Therefore, this is an area that will benefit from
assert power, record is for showing what has happened and research into the transition around information gathering
is happening, while information for action means that those and use changes.
who are required to act get the information they need. The This leads to the following question: how may data
principle of information flow is used and further developed be gathered and used for information when organizations
by several authors. For instance Curşeu et al. (2021) dis- strive to become data-driven?
cuss how unequal power may be balanced with information Norway ranks fifth on the European Digital Economy
flow, and Lin and Cornford (2000) to explain the connection and Society Index (DESI), behind Finland, Denmark, the

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
68 K. Wulff, H. Finnestrand

Netherlands and Sweden. The DESI index measures the sultancy. Because the research we had found was all in-
level of a country’s digitalization. The finance and public ternational, we decided that it would be important to get
sectors are two front runners in the area of digitalization. a better grip on the situation in Norway. This means that our
The public sector has been inspired by the governmental sampling strategy was purposive (Miles et al. 2018). The
strategy of “one digital public sector” (Norwegian Govern- purpose we followed when sampling was to find the people
ment 2019) while finance is leading the way in becoming that were most experienced in the area of data and ML use.
more data-driven (Almquist et al. 2021), assisted by the The AI research group collaboratively designed the inquiry
fact that a large part of their offerings can be produced and (Shani and Coghlan 2021). We carried out interviews on
delivered digitally. In this study, we have interviewed ex- topics related to data and ML with experts mainly from the
perts and leaders from organizations becoming data-driven banking and finance and public sectors—finance because
in both sectors. Their organizations are on different stages it is one of the leading sectors (Almquist et al. 2021), and
of becoming data-driven: some are only just starting out, the public sector because of the government’s requirements
others have been trying to transform, but without much to use and share data (Norwegian Government 2019). We
success, while the third kind are ahead of the general devel- interviewed both front runners and organizations that were
opment and already reaping huge benefits from their data. at the beginning of their journey.
In this explorative study, we have collected viewpoints from We conducted a total of 24 expert interviews across
the experts on the state of the development in their organi- 13 organizations in the period between May 2020 and
zation through retrospective interviews. March 2021 (Table 1). The interview guide for the quali-
tative, semi-structured interviews was iteratively and col-
laboratively developed by the group doing the interviews
2 Research design (Holstein and Gubrium 1995). Because we wanted to have
rich context descriptions, we decided to have little prior
The empirical data for this article is based on a research instrumentation (Miles et al. 2018). The interviews were
project conducted by a group of nine data professionals open-ended and explorative (Gudmundsdottir 1996) and
and leaders, including the first author, from a software con-

Table 1 Overview of the organizations and interviewees in the study

Nr. Role Organization Approx. org. size Sector Type of business
1 Development Director Alvarinho 100 Private Software products
2 Senior Vice President (VP) Barbera 9000 Private Finance
3 Head of Robotics & AI Barbera 9000 Private Finance
4 Chief Data & Analytics Officer (CDAO) Barbera 9000 Private Finance
5 Programme Leader Anti-Money Laundering Carignan 7500 Private Finance
(AML)
6 IT Developer Carignan 7500 Private Finance
7 Chief Information Officer Dolcetto 400 Public Directorate
8 Credit Manager Egiodola 50 Private Finance
9 Solution Architect Egiodola 50 Private Finance
10 Head of Innovation Egiodola 50 Private Finance
11 Chief Data Officer (CDO) Fiano 900 Private Finance
12 Head of Automation and Business Analytics Fiano 900 Private Finance
13 Head of Machine Learning and AI Fiano 900 Private Finance
14 Head of Architecture and Platform Garnacha 250 Private Finance
15 Product Owner Data Platform Hamashara 6600 Public Directorate
16 Head of Automation Hamashara 6600 Public Directorate
17 Director Impigno 2600 Private Finance
18 Senior Vice President IT Impigno 2600 Private Finance
19 Senior Enterprise Architect Impigno 2600 Private Finance
20 Analysis Leader Jampal 550 Public Energy provider
21 Chief Technology Officer Katsano 1250 Private Finance
22 Strategic Advisor Lambrusca 1500 Private Finance
23 Executive Director Lambrusca 1500 Private Finance
24 Digital Product Manager Marsanne 550 Private Maritime technology

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Data-driven information for action 69

we built on the knowledge we gathered and changed the logic was coherent, in what Lincoln and Guba (1985) call
next interview accordingly. a confirmability audit. They suggested a two-level structure,
The main focus of the questions was to understand how which was implemented by the first author. Then, the au-
top management was involved, what roles were defined thors agreed on which codes to research further, thus engag-
in top management and other levels of the organization, ing in an inductive-deductive process. To get an overview
how the ML work was organized, how the interaction be- of the quotes in the different codes, the first author printed,
tween the design team and the rest of the organization was cut out, and sorted the quotes and divided those codes that
handled, what experiences and challenges the interviewees had become too broad into subcodes (Miles et al. 2018)
meant that their organization had had with regards to be- before entering them into a mindmap. The mindmap was
coming more data-driven, and what areas the interviewees then studied by the first author and the research group, and
saw as having benefitted from using ML. patterns of positive and negative influence pointed out and
Of the 24 interviews, 22 were audio recorded. The discussed. These were described in causal fragments, be-
recording was done by connecting an audio recorder to fore joining them into two causal networks (Miles et al.
the MS Teams meeting. The transcription was done verba- 2018). One causal network depicted the actions taken by
tim, i.e., everything that was said, including “ehms” and organizations that were very successful in their work to be-
such, was written down (Poland 1995), without noting any come more data-driven, and the other the actions taken by
nonverbal communication. Two of the interviewees were the less advanced organizations. The causal networks were
Swedish, and those were translated directly to Norwe- tested on three colleagues outside the AI research group in
gian in the transcription, while the two English interviews a peer debrief (Lincoln and Guba 2016). These discussions
were transcribed in English. Once the transcription was and visualizations provided a foundation for selecting an
approved, the audio recording was deleted. Due to COVID- area of difference between the organizations and helped the
19, 23 of the 24 interviews were conducted online on MS authors define the area of analysis. The selected area was in-
Teams. The questions and a consent form were approved formation flow and its three variants: information for record,
by the Norwegian Centre for Research Data (NSD). action, and control. Thereafter, the selected codes (quirks)
A thorough first cycle coding was performed as the tran- and quotes were analyzed to find illustrative examples to
scribed interviews were coded in Quirkos with the use of present here. The quotes in Norwegian were translated by
in vivo codes, concept codes and process codes (Miles the first author and double-checked by the second author.
et al. 2018). The interviews were analyzed using theme- According to Bulling (2018), organizational change can
based analysis with a focus on the content of the narrative, be described by its degree of change, how broad the em-
as opposed to a structural analysis focusing on how the ployee involvement is, and who leads the change. The AI
story is told (Riessman 2008). The coding was validated by research group used this categorization to evaluate the or-
two representatives from the research group, contributing ganizations. In addition to the three factors described by
to the confirmability of the selection to ascertain that the Bulling, we have added information from the interviews on

Table 2 Evaluation with regards to becoming data-driven based on Bulling (2018) and status given in the interviews
Degree of organizational change (Bulling 2018) Status in the organizations
Organization Degree of Degree of involvement Who is leading the change Active use of ML Solutions in produc-
change since tion
Alvarinho Incremental Several groups Strategic management 2018 Yes
Barbera Incremental The whole organization Strategic management 2019 Yes
Carignan Incremental One team The developers 2018 Yes
Dolcetto Incremental Single persons Middle management 2020 No
Egiodola Incremental Single persons Middle management 2013 No
Fiano Incremental Several groups Strategic management 2018 Yes
Garnacha Incremental One team Middle management 2020 No
Hamashara Incremental Single persons + one The analysts and developers 2019 Yes
team
Impigno Incremental One team Strategic management 2017 Yes
Jampal Incremental One team Analysts 2017 Yes
Katsano Incremental Several groups Strategic management 2016 Yes
Lambrusca Incremental Single persons Strategic management 2020 No
Marsanne Radical Several groups Business and digital devel- 2016 Yes
opers

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
70 K. Wulff, H. Finnestrand

when the organizations started using ML actively (that is for and therefore stopped recording it at some point. This
with the intention for using it in production) and whether is illustrated in the following quote:
any of their solutions had reached production at the time of
Since we have a very low awareness on what we can
the interview.
get from the data, we have not been good at taking
As can be seen from Table 2, most of the organizations’
care of our data. We still write information that is
degree of organizational change so far had been incremen-
important for operations down by hand and store it in
tal, for instance by creating ML models to detect fraud or
ring binders in people’s offices. It is incredible. For
reduce customer churn. This can also be seen from the dif-
the operation of the power plants for instance, there
ferent examples we will present next. In the following we
has been no trend analytics [predictive analytics] on
have chosen to focus our analysis on two of the three vari-
any of them. Here, data was noted in books, and at
ants of information flow: information for record and for
some point, they just stopped recording it. And no-
action. This is partly because the third variant, information
one restartet doing it, because they didn’t know what
for control, is something Cherns (1987) advises against, and
to do with it. (Analysis leader, Jampal)
partly because we believe that control issues will have to
be researched in another way, for instance, by observing The fact that someone stops recording the data because
how the information flow is actually designed. We start our they do not know what to do with it, points to a process
exploration with the first contact with the data—the data where there is no focus on information for action. The data
gathering. that was recorded in binders may have been information for
action at one point, but storing it in binders is not consistent
with using the data for learning purposes as it is hard to
3 The empirical perspective—what the compare trends and do other analyses on such data.
experts say How to store data was seen by some as a personal choice,
and others’ need for the same data was not seen or acknowl-
3.1 Data gathering edged. Some informants told us about professionals who re-
garded their data and models as their personal intellectual
The informants were aware of the new possibilities for property, as illustrated in this quote:
designing information that emerges from recording data
You have some professional areas that can be con-
through digitalization. However, the organizational aware-
sidered half art and half engineering. For instance,
ness differed. Some of them were particularly conscious
detecting quick clay through taking drill core sam-
about what to store and what not, while others struggled
ples. Then you get a pressure profile and can measure
with getting data recorded in a retrievable way.
water penetration and guess whether there is quick
Now, we are much more aware on how we gather data clay present or not. And this is not an exact science,
and store it from all kinds of customer processes and it is intuition and such. The mathematical models that
business processes. That gives us a totally different people use are not efficient, it is 50/50. It seems that
set of data than we had earlier. Before, you had to people saw this as their interpreations, it was their
access many different sources to find and gather the intellectual property. So, when we began doing ma-
data. It is still a challenge to group all the data before chine learning on this many protested against sharing
one begins the processing, but much of the data is their data, but after a while we gathered enough data
already structured so that it is easy to process without to run predictions and found 97% correlation between
needing to adapt it to for instance a machine learning the drill core samples and the more precise method of
model. (Senior VP, Barbera) core drilling. (Head of Automation, Hamashara)
In this quote from Barbera, we can see how important it By calling it half art/half engineering, the informant illus-
is that the data is easily accessible, for example for build- trates how the ownership of the data is important to each
ing predictive ML models. Barbera reported that they had individual. This makes it understandable that it is hard to
been quite conscious and explicit about decisions on what share the data. Therefore, it seems important to find ways
to record from the different processes. However, other or- of increasing individual awareness that the data gathered
ganizations had not reflected about it in the same way. can be used for collective learning purposes.
They were not digitalized to the same degree, and digi-
talization was hampered by differences in how the data
was recorded. For instance, those organizations realized
that people recorded data for their own actions only, or
that they recorded data without knowing what it was used

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Data-driven information for action 71

3.2 Data use the usefulness of data in advance. In another case, the or-
ganization discovered that they wanted data that they did
All of the informants were passionate about using the data not have. Because of COVID-19, some organizations expe-
available to inform action, as illustrated in the following rienced an unprecedented increase in uncertainty. For the
quote: leaders of these organizations, there was an increased need
for information to give insight into possible strategic ac-
What I want is that it [information] should give us bet-
tions. For instance, a board of directors was mentioned as
ter insight. In a way it should add brainpower to the
a group needing more information to handle the situation.
people analyzing data and help us to see patterns in
data to pick up changes early. For instance, changes in We have an interesting case now, because corona has
customer behavior or other things, changes that hap- done things with our business that is impossible to
pens around other actors. We see it as an added pro- simulate. For instance, at the worst time, our revenue
cessing power that is connected, that fits what people halved. So, now we have a board that jumps up and
are looking for themselves already. (Solution archi- down and wonders, ‘We must have estimates, what
tect, Egiodola) will this mean for the result this year, and what will it
mean for next year?’ They think we are fortune tellers.
However, the passion of the informants sometimes turned
We do not know what will happen. Will people be
into frustration when they met road blocks in the organiza-
able to afford to pay the same as before? (Solution
tional design that reduced the organizations ability to make
Architect, Egiodola)
use of the data.
The situation that the board of directors—and many oth-
I believe this is an awareness thing. If you look at
ers—were in, illustrates the limits of being data-driven.
the value chain, the “data gang” or the people having
Without data from similar situations, it can be impossi-
the data mindset sit at the end of the value chain. If
ble to do predictions. Therefore, to have data that shows
you want to get the optimal setting you need to move
what is happening in real-time can be beneficial. This was
that mindset to the front of the value chain where the
explained to us by two of our informants. These two orga-
customer is, and the transaction occurs. This could
nizations shared data with the Norwegian government and
lead to an awareness for the people on the front desk to
Statistics Norway to support decisions in connection with
understand that all the information about the customer
COVID-19.
or the process has great value later in the process, and
that the quality of this data needs to be perfect. Then Many governmental actors like for instance Statistics
I think we can succeed. But it will take time. What Norway use data from the payment infrastructure in
many companies do is that they improve the low data their models to increase the quality of macro-econom-
quality afterwards. That is not a good choice, you ical decisions. Delivering such data is part of our per-
need to be able to improve the front of the value chain. mission to run this infrastructure. One example where
(Senior VP, Barbera) the insight from the infrastructure has been used for
big and fundamental decisions for Norway was when
One informant pointed out that information which was built
a committee evaluated the work done in connection
for use in the ML model was discovered to have other uses
with COVID-19 and recommended how to proceed.
for action as well. For example, one of the organizations
In such macro-decisions, the possibility to have easy
discovered that the data that was fed to the ML model for
access, continuously and preferably in real-time, is
identifying any possible anti-money-laundering cases was
essential (Director, Impigno)
so valuable in itself, that they chose to make it available to
the people working with such cases. This means that the necessary data may be shared both
inside and between organizations.
We began with a simple supervised ML model. We
realized then that the data we use to train the model
3.3 Data sharing
gives a lot of value to the banks as it is. There is
a lot of data that is relevant when you are working on
Public organizations in Norway are required to share data
anti-money-laundering cases, but that the case work-
outside their organization (data.norge.no). This led one of
ers have not had access to. So, in addition to the ML
the public organizations to state that “data that is not ex-
project we have built and made available a report with
posed is not real” (Head of Automation, Hamashara). To
the same data. (Program Leader AML, Carignan)
expose data is to make it available for others to extract. An-
Here, the expectations of what the data would be used for other public organization (Dolcetto) mentioned how they
expanded, which means that one cannot always determine shared data with public organizations for real-time use of

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
72 K. Wulff, H. Finnestrand

the data in those organizations’ software solutions. In this from, for instance data warehouses and buy business intel-
way they contribute to better services from different public ligence (BI) solutions, for the cloud-based data platforms
organizations, and probably also better coordination. One of today, one needs algorithm and ML competence to ex-
informant told us about a gas station chain that asked for tract and create a useful presentation of the data. Because
access to traffic pattern data to better estimate the number such competence was hard to find, one of the organizations
of buns to warm up, thereby reducing the waste of food. In upskilled their employees:
this way, the data was intended to support another organi-
What we saw was that we needed more data scientists,
zation’s actions. The data shared with government was to
and it is hard to find people to employ. So, then we
provide the foundation for decisions that had to be made,
established a reskill program. There, we educated ap-
which function as information for action. Private organi-
proximately 30 people over a year as data scientists.
zations also shared data, both internally between business
This was done through a cooperation with universi-
areas (Marsanne) and externally: one of the banks asked
ties and through practical cases and work. (Senior VP,
the customers’ consent for using their transaction data to
Barbera)
inspire them to reduce their CO2 footprint.
Some of the informants explained how access to data had In addition to such upskilling, some of the informants also
been a discussion point in their organization. Some organi- emphasized the importance of understanding the data and
zations had previously had a strict governance framework what you could actually achieve from using the data, as
for the data, partly because of legal regulations and partly well as the possibility of supporting the learning process
for security reasons. This was challenged due to new data with data:
platforms that made it possible to share data on a wider
I believe that the area where we will see development
scale. From there, they could proceed in one of two direc-
now is that we will become better at using available
tions: either try and limit control to a few who were allowed
data sources. We will develop a better gut-feeling on
to access the data, or to give wide access rights and trust the
what data that actually informs us, and what is just
people to make good decision. However, there is a need for
nonsense. (Head of Innovation, Egiodola)
understanding what data can be shared and what cannot:
You are going to democratize your data, but you really
can’t do that if you don’t understand what data you
4 Discussion
can democratize. Are people really allowed to use it?
(CDO, Fiano)
Our research question is: how may data be gathered and
By democratizing the data, the idea is to let everyone decide used for information when organizations strive to become
for themselves whether the data will give them the infor- data-driven?
mation they need or not. This can lead to changes in the When becoming data-driven, an early step is to under-
principles for access management. Some informants gave stand what data is available and what data will need to be
access based on skills: recorded. This may be data that was previously documented
as information for record, that is, where the data was reg-
The data itself is managed within our multi cloud so-
istered without a direct use. Such data may be valuable
lution. From a data perspective the cloud platforms
now, but the way of recording the data may not have been
provide highly secure, massively scalable and agile
updated to a digital medium, as in the example of Jampal
delivery solutions. The cornerstone of our enterprise
where some of the information was recorded in binders.
data and analytics platform, has been the establish-
We believe this to be a sign of a low-learning organization,
ment of our advanced cloud-based data science labo-
and to change this behavior one would need to facilitate
ratory which provides the analytics capability and the
changes in the local work process. To provide the organiza-
technical scalability to over 40 data scientists across
tion with data literacy competence as suggested by Berndts-
the organization. (CDAO, Barbera)
son and Svahn (2020), may help somewhat, but more im-
As pointed out, here only the data scientists had access portant would be to help the organization with redesigning
to the data. Some of the organizations told us that they the structure and process so that the data that is recorded
were aiming to build data products, that is, refined data sets provides meaning for it and opens it up for learning. The
combined with an application that processed the data and Barbera example, in contrast, showed an organization that
generated results, made easily available to a larger part of is very engaged with its data and oriented towards making
the organization. This because the competence needed to it more easily available to enable learning. In a STSD per-
extract and use the data was an obstacle. While the earlier spective we would expect that the processes that produce
data storages was somewhat easier to extract information the data are designed in a way so that the recording of the

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Data-driven information for action 73

data benefits the group that does the recording, for instance, of the data. It also means that the exploration on what data
by becoming their information for action. This may be done to record should have wide boundaries.
by creating their own interfaces, in a way consistent with The intention behind becoming more data-driven is to
the Archipelago architecture described by (Govers and van improve the efficiency of the work processes either through
Amelsvoort 2018). augmentation (supporting the worker) or automatization, to
With regards to information for record we also saw that improve the customer experience, create new business mod-
in some of the organizations there was individual ownership els and to better support decision-making (Benbya et al.
of the data and an unwillingness to share. As described in 2020). When it comes to information for action, it is sup-
Gieryn (1983), the individual’s behavior may be intended posed to aid the team doing the work. When the task given
to protect the worker’s professional field, and as Wulff and is to improve and automatize the work, the necessary com-
Finnestrand (2022) illustrated, this may well be because petences needed may be so different from the competence
there is no alignment process that supports a more collab- to perform the work that other experts are invited in, for
orative work environment. It may also be seen as a power instance programmers. This creates a distance between the
issue—to have information is to have power—and if this is people performing the work as it is today and the people
the case we would expect to find structures in the organi- changing the work that should be kept as small as possible
zation that promotes such behavior. For instance, that the (Orlikowski 1992).
quick clay experts are acknowledged for being sought after, When the delivery is fully digital, as is the case for many
that is, which creates a competition around who is the best of the processes in finance, the design and development
expert. To alleviate such problems, the experts could be re- team is doing the work, and can, with feedback try to con-
organized into an organizational structure with a common tinuously improve the deliveries. This means that much of
purpose to improve the detection of quick clay. This could the information for action when becoming data-driven is
help the transition from locally recorded data to making the design-related. And the design includes the output of the
data available for the collective, and to ensure that the data team. One of the most notable differences between the work
is made available and that it is used by the people needing teams in British coal mines, which have informed most so-
the data to improve the way they work. ciotechnical design studies (Trist and Bamforth 1951), and
That data is recorded by people may become less com- digital product development teams nowadays, is probably
mon because many processes are digitalized and thereby that coal miners were never encouraged to decide for them-
provide data automatically. In addition sensors and other selves whether they should go looking for other products.
hardware are providing data. Nevertheless, the decision on They could not one day decide to go looking for gold, or
what to record lies with the people designing the digital emeralds, or to find out if it was possible to extract valu-
solutions, and to do so demands a knowledge of the work able minerals from the coal. Digital product development
done that can only come from participation of the workers teams—like many other teams today—are constantly asked
that perform the job. For instance, that the people doing the to create new products or services. As a result, information
maintenance work are involved in gathering the data to find for action is not only necessary for employees in order to
improvements. perform daily operations, but also in order to be able to
Although recorded data may not be used for action in be innovative and discover new products or services. Then,
the organizational unit that records it, it is often useful for information for action is seen as a way to try to better un-
another business area, another organization, or perhaps for derstand what is going on in and around the organization, to
future use. The Senior VP from Barbera explained how the be a learning organization (Herbst 1974, 1993). The differ-
data quality is created in another part of the value chain ence between an organization that has a limited interest in
than the one using it. This means that a part of data quality learning from the data and an organization that is well un-
lies in organizational design. When the use of the data is in derway to benefitting from its data have been described in
another area of the organization—or even outside of it—it the previous chapter. We would like to see even more orga-
challenges the idea of information for action being steered nizational design research on how to help the low-learning
back to the group doing the action so that they can evalu- organizations develop.
ate their actions and find new and better ways to achieve We found that the use of data to inform action is hap-
the intended outcome (Herbst 1974). This means that the pening in the primary processes as well as in tactical and
organization can try to find mechanisms that will allow the strategic processes, which aligns with the findings of Halper
group recording the data enough insight into the data quality and Stodder (2017). This means that when designing the in-
the other group needs for their information for action. Per- formation that will lead to action there is a need to reflect
haps there could be some kind of transparent feedback loop upon all types of use, not only the use taking place in pri-
between the group recording the data and the one needing mary processes. What we also found was that data can be
it for their action. This may create pride in the usefulness

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
74 K. Wulff, H. Finnestrand

used for other purposes than intended, which means that mation for action and what can be considered information
there is a need for openness in the access to the data. for control.
Cherns (1976, 1987) addressed work designers of the
work in his seminal articles, but now the possibilities may
also increase for the workers to design their information for 5 Conclusion and considerations for practice
action for themselves. The quality of work as described by
de Sitter et al. (1997) implies that the workers are given Our study of 13 different organizations at different stages
sufficient autonomy to handle the variance they experience. in their quest to become data-driven points to 4 findings to
As the variance they experience increases, the need for in- guide us.
formation to understand what is going on will probably First, when there is a need to record data that is not di-
increase as well. This is why one element of data shar- rectly used as information for action by the people doing
ing, the possibility to enable your own information through the recording, there may be challenges with regards to how
design might be useful. The new data platforms provide well the data is gathered. We suggest that more research is
possibilities for using data in new and innovative ways. We done to find ways of handling this, for instance by mak-
saw that one of the experts from Fiano believed in the de- ing the actions supported by the data data transparent. To
mocratization of data, which means that people should be be able to utilize the opportunities that lie in being a data-
able to create their own information for action. However, driven organization, it is important to gather and system-
this also affects design choices. Some of the interviewees atize the information in such a way that it becomes joint
believed that the access to the data should be restricted. information for action.
This may be because they want to limit perceived vulner- Secondly, for organizations where part or all of the de-
abilities that appear if everyone can access data and build livery is digital, the information for action is often design-
their own reports on what is going on. Or, because data ac- related. That is, it is used to redesign the digital processes.
cess previously may have been restricted to the users of one There is a danger that the development team will wish to
particular IT-solution, this may be the way the access man- avoid going into organizational development issues where
agement has been defined. Now, there can be discussions only parts of the process are digital, like in the Jampal ex-
on who will have access and when. This is why Davenport ample. This is similar to what was described by Smith and
and Kirby (2016) explain that data analysts should make an Eckroth (2017), whose suggestion is to avoid affecting ex-
effort to become data scientists, that is, programming the isting work process when designing AI solutions. To design
reports into ML models instead of using a lot of time to such AI solutions can lead to investments into AI diverting
gather data and run analyses. from what is most beneficial to the organizations. There-
As demonstrated by Mitki et al. (2019), providing the fore, there is a need to understand what task structures can
workers in the organization with information on how the and should be redesigned and where AI can contribute.
organization is performing is a powerful improvement tool. Third, and building on this, becoming data-driven while
We believe such a sharing also to be part of the idea in the still maintaining a flexible organization (Brödner and Lat-
democratization projects performed by Emery and Thorsrud niak 2003) means that the information that is important for
(1976), although then only in the business unit they were action today may not be what is important tomorrow. To en-
a part of. However, in some cases information is not used for sure that the digital transformation results in a more flexible
action but for extraneous reasons, such as curiosity, which organization, we think it is essential to enable more people
is why banks monitor who accesses a customer account as in the organization to participate in deciding what data to
instructed by law. Such extraneous uses may be used to record in order for them to come up with new information
justify why the data access should be limited, however, we and new uses for it. This means that people without prior
believe that the information for action needs should take knowledge of designing information for action are now ex-
precedence. pected to do this. To support their advances into building
When it comes to improving the decision-making in the information for action, the data must not only be available,
organization, STSD is clear on the principle that control but the person must also have the required competence to
is built bottom-up. This is why information for action is build the information in compliance with the regulations.
steered to the group/team doing the work. This means that Fourth, the need for information for action is spread
the design principles for information for action for strategic across the whole organization. This means that it will differ
purposes, needed for instance by of a board of directors or where and what information is needed first, and that it is
by top management, are difficult to deduce. As the possi- important to share it for the right purpose regardless of the
bilities for information for control increases with AI, we person’s position. The information for action is, in addition
would like to encourage research into this area to provide to its use for operational action, also used to understand
examples or guidelines of what constitutes strategic infor- what is happening in the outside world. It is needed on

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Data-driven information for action 75

several levels in the organization, by staff in operations, colleagues in the AI research group who have contributed to the
data collection and learning, especially Edvard K. Karlsen and Frode
in design, and in leadership positions. This means that the
Standal who have been vital in this research. A warm thanks also to
boundaries of what can be categorized as information for the reviewers that have provided us with many insightful comments to
action should be explored and reflected upon by the people help us improve this paper. This research is supported by The Research
involved in the redesign. Council of Norway under Grant number 290629 in an Industrial PhD
project for the software company Kantega AS.
Our suggestion for future research is into the area of
data design and organizing. We believe that now, perhaps Funding Open access funding provided by NTNU Norwegian Univer-
more than ever, as the foundations for tomorrow’s data- sity of Science and Technology (incl St. Olavs Hospital - Trondheim
driven businesses are being laid, there is a need for a strong University Hospital)
presence from STS researchers and consultants. Therefore, Open Access This article is licensed under a Creative Commons At-
it is important to engage in reflections on what data to tribution 4.0 International License, which permits use, sharing, adapta-
record in order to build quality organizations. Not every tion, distribution and reproduction in any medium or format, as long as
you give appropriate credit to the original author(s) and the source, pro-
data element that is produced can and should be stored, and vide a link to the Creative Commons licence, and indicate if changes
the discussion on what data to use for action may inform were made. The images or other third party material in this article are
the discussions about digitalization of the work. included in the article’s Creative Commons licence, unless indicated
Finally, becoming data-driven is sociotechnical in na- otherwise in a credit line to the material. If material is not included
in the article’s Creative Commons licence and your intended use is not
ture, and the major difficulties are nontechnical (Halper and permitted by statutory regulation or exceeds the permitted use, you will
Stodder 2017). Organizations are now hiring—or trying to need to obtain permission directly from the copyright holder. To view
hire—data scientists to handle their data and to seek value a copy of this licence, visit http://creativecommons.org/licenses/by/4.
from it (Benbya et al. 2020). In addition we would encour- 0/.
age that they train or hire organizational designers. As we
have pointed out, there are areas where there is benefits in References
upskilling or reskilling employees, and it could be bene-
ficial to increase workers’ design knowledge (Parker and Almquist, A., Gjengstø, M., & Nordli, J. A. (2021). Data-driven Busi-
ness. Application of data in Norwegian business and how to
Grote 2022) so that they are able to design and use the data transform a data-driven organisation. https://www.itera.com/
for action. datadriven-2021#download-ddb-report. Accessed 24 Aug 2021.
At the same time that we interviewed the experts, another Argyris, C., & Schon, D. A. (1974). Theory in practice: Increasing pro-
software consultancy in Norway made a quantitative survey fessional effectiveness. San Fransisco: Jossey-Bass.
Babüroğlu, O. N., & Selsky, J. W. (2021). Toward Reconfiguring
of the situation in Norway with regards to becoming data- Sociotechnical systems design: digitally infused work sys-
driven. One of their findings was formulated as spread- tems and the “platform-STS”. In A. B. Shani & D. A. Noumair
sheets are (still) king (Almquist et al. 2021, p. 7) which (Eds.), Research in organizational change and development
points to the same individual use of data that we found. (pp. 63–87). Bingley: Emerald. https://doi.org/10.1108/S0897-
301620210000029004.
They also point out that where leaders have a clear ac- Bean, R. (2021). Fail fast, learn faster. Lessons in data-driven lead-
countability for making data available, it is democratized. ership in an age of disruption, big data and AI. Hoboken: John
This means that the need for joint information for action Wiley & Sons Inc.
still waits to be discovered in some organizations, and that Benbya, H., Davenport, T. H., & Pachidi, S. (2020). Special issue ed-
itorial: Artificial intelligence in organizations: current state and
leaders will need to both make the data available and see future opportunities. MIS Quarterly Executive. https://doi.org/10.
to it that the people in the organization can and will use it. 2139/ssrn.3741983.
Summing up, the short answer to our research question Berndtsson, M., & Svahn, T. (2020). Strategies for scaling analytics:
of how data is gathered and used when becoming more data- A nontechnical perspective. Business Intelligence Journal, 25(1),
43–53.
driven is: not easily. As we have illustrated there are orga- Berndtsson, M., Forsberg, D., Stein, D., & Svahn, T. (2018). Becom-
nizational design roadblocks to handle, for instance where ing a data-driven organisation. In 26th European Conference on
the organizational design has not supported mutual learning Information Systems (ECIS2018), Portsmouth, United Kingdom.
so far, or where the process is divided so that the people Boonstra, A., & Reezigt, C. (2019). Complexity-predictability project
diagnosis model. Procedia Computer Science, 164, 337–342.
expected to gather the data are not the ones reaping the https://doi.org/10.1016/j.procs.2019.12.191.
benefits. Then again, we have seen examples of organiza- Brock, J. K.-U., & von Wangenheim, F. (2019). Demystifying AI:
tions that have managed to provide their employees with the what digital transformation leaders Can teach you about realis-
room and the competence needed to progress the organi- tic artificial intelligence. California Management Review, 61(4),
110–134. https://doi.org/10.1177/1536504219865226.
zation. This, hopefully, will provide inspiration to others Brödner, P., & Latniak, E. (2003). Sources of innovation and compet-
trying to become more data-driven. itiveness: National programmes supporting the development of
work organisation. Concepts and Transformation, 8(2), 179–211.
Acknowledgements We would like to thank the informants who https://doi.org/10.1075/cat.8.2.06bro.
shared all their valuable experience and knowledge with us, and our

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
76 K. Wulff, H. Finnestrand

Brynjolfsson, E., & McAfee, A. (2014). The second machine age: Herbst, D. P. G. (1974). Designing with minimal critical specifications.
Work, progress, and prosperity in a time of brilliant technologies. In P. G. Herbst (Ed.), Socio-technical design: Strategies in multi-
New York: WW Norton. disciplinary research (pp. 294–302). London: Tavistock.
Brynjolfsson, E., & McElheran, K. (2016). The rapid adoption of data- Herbst, D. P. G. (1993). A learning organization in practice, M/S Balao.
driven decision-making. American Economic Review, 106(5), In E. Trist & H. Murray (Eds.), The social engagement of social
133–139. science: a Tavistock anthology (pp. 409–416). Philadelphia: Uni-
Bulling, K. (2018). The systems constellation as an instrument for versity of Pennsylvania Press.
change agents. Bremen: Springer Gabler. https://doi.org/10.1007/ Holstein, J. A., & Gubrium, J. F. (1995). The active interview. Vol. 37.
978-3-658-24908-3. SAGE.
Cherns, A. (1976). The principles of sociotechnical design. Human Re- Iansiti, M., & Lakhani, K. R. (2020). Competing in the age of AI, strat-
lations, 29, 783–792. egy and leadership when algorithms and networks run the world.
Cherns, A. (1987). Principles of sociotechnical design revisited. Hu- Boston: Harvard Business Review Press.
man Relations, 40(3), 153–162. https://doi.org/10.1177/00187267 Janssen, M., van der Voort, H., & Wahyudi, A. (2017). Factors in-
8704000303. fluencing big data decision-making quality. Journal of Business
Claussen, T., Haga, T., & Ravn, J. E. (2019). Socio-technics and be- Research, 70, 338–345. https://doi.org/10.1016/j.jbusres.2016.08.
yond: an approach to organisation studies and design in the second 007.
machine age. European Journal of Workplace Innovation. https:// Knight, C., & Parker, S. K. (2021). How work redesign interventions
doi.org/10.46364/ejwi.v4i2.569. affect performance: An evidence-based model from a systematic
Curşeu, P. L., Semeijn, J. H., & Nikolova, I. (2021). Career challenges review. Human Relations, 74(1), 69–104. https://doi.org/10.1177/
in smart cities: A sociotechnical systems view on sustainable ca- 0018726719865604.
reers. Human Relations, 74(5), 656–677. https://doi.org/10.1177/ Lin, A., & Cornford, T. (2000). Sociotechnical perspectives on emer-
0018726720949925. gence phenomena. In E. Coakes, D. Willis & R. Lloyd-Jones
Davenport, T. H., & Kirby, J. (2016). Only humans need apply: Win- (Eds.), The new sociotech (pp. 51–60). London: Springer.
ners and losers in the age of smart machines. New York: Harper Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Beverly
Business. Hills: SAGE.
Davenport, T. H., Barth, P., & Bean, R. (2012). How ‘big data’ is dif- Lincoln, Y. S., & Guba, E. G. (2016). The constructivist credo. London:
ferent. MIT Sloan Management Review, 54(1), 22–24. Routledge. https://doi.org/10.4324/9781315418810-3.
Delen, D., & Ram, S. (2018). Research challenges and opportunities Makarius, E. E., Mukherjee, D., Fox, J. D., & Fox, A. K. (2020). Rising
in business analytics. Journal of Business Analytics, 1(1), 2–12. with the machines: A sociotechnical framework for bringing arti-
https://doi.org/10.1080/2573234X.2018.1507324. ficial intelligence into the organization. Journal of Business Re-
Emery, F., & Thorsrud, E. (1976). Democracy at work. The report search, 120, 262–273. https://doi.org/10.1016/j.jbusres.2020.07.
of the Norwegian industrial democracy program. Vol. 2. Leiden: 045.
Martinus Nijhoff Social Sciences Division. Meacham, B. J. (2022). A sociotechnical systems framework for per-
Faraj, S., Pachidi, S., & Sayegh, K. (2018). Working and organizing in formance-based design for fire safety. Fire Technology. https://
the age of the learning algorithm. Information and Organization, doi.org/10.1007/s10694-022-01219-0.
28(1), 62–70. https://doi.org/10.1016/j.infoandorg.2018.02.005. Mikalef, P., & Krogstie, J. (2020). Examining the interplay between big
Galbraith, J. R. (2014). Organizational design challenges resulting data analytics and contextual factors in driving process innova-
from big data. Journal of Organization Design, 3(1), 2–13. tion capabilities. European Journal of Information Systems, 29(3),
Gieryn, T. F. (1983). Boundary-work and the demarcation of science 260–287. https://doi.org/10.1080/0960085X.2020.1740618.
from non-science: Strains and interests in professional ideolo- Miles, M. B., Huberman, A. M., & Saldaña, J. (2018). Qualitative data
gies of scientists. American Sociological Review, 48(6), 781–795. analysis: A methods sourcebook (4th edn.). SAGE.
https://doi.org/10.2307/2095325. Mitki, Y., Shani, A. B., & Greenbaum, B. E. (2019). Developing new
Gobble, M. M. (2018). Digitalization, digitization, and innovation. Re- capabilities: a longitudinal study of sociotechnical system re-
search-Technology Management, 61(4), 56–59. https://doi.org/10. design. Journal of Change Management, 19(3), 167–182.
1080/08956308.2018.1471280. Nerur, S., Cannon, A., Balijepally, V., & Bond, P. (2010). Towards an
Govers, M., & van Amelsvoort, P. (2018). A socio-technical perspec- understanding of the conceptual underpinnings of agile develop-
tive on the design of IT architectures: The lowlands lens. Manage- ment methodologies. In T. Dingsøyr, T. Dybå & N. B. Moe (Eds.),
ment Studies, 6(3), 177–187. https://doi.org/10.17265/2328-2185/ Agile software development (pp. 15–29). Heidelberg: Springer.
2018.03.003. Norwegian Government (2019). One digital public sector. Digital strat-
Govers, M., & Südmeier, P. (2016). Applying enterprise information egy for the public sector 2019–2025. https://www.regjeringen.no/
technology from a socio-technical perspective. In Co-creating hu- en/dokumenter/one-digital-public-sector/id2653874/. Accessed 1
mane and innovative organizations: Evolutions in the practice Nov 2022.
of socio-technical system design (pp. 289–302). Portland: Global Orlikowski, W. J. (1992). The duality of technology: Rethinking the
STS-D Network. concept of technology in organizations. Organization Science,
Gudmundsdottir, S. (1996). The teller, the tale, and the one being told: 3(3), 301–441.
the narrative nature of the research interview. Curriculum Inquiry, Osmundsen, K., Iden, J., & Bygstad, B. (2018). Digital transforma-
26(3), 293–306. tion: drivers, success factors, and implications. In MCIS (Mediter-
Haga, T. (2019). Can an unconventional socio-technical system ap- ranean Conference on Information Systems), Corfu, Greece.
proach open the way to new solutions and new understanding? Parent-Rocheleau, X., & Parker, S. K. (2022). Algorithms as work
European Journal of Workplace Innovation. https://doi.org/10. designers: How algorithmic management influences the design
46364/ejwi.v4i2.567. of jobs. Human resource management review. https://doi.org/10.
Halper, F., & Stodder, D. (2017). What it takes to be data-driven. 1016/j.hrmr.2021.100838.
(TDWI Best Practices Report, December, Issue). https://media. Parker, S. K., & Grote, G. (2022). Automation, algorithms, and be-
bitpipe.com/io_14x/io_141315/item_1674359/TDWI_BPReport_ yond: Why work design matters more than ever in a digital world.
Q417.pdf. Accessed 7 November 2022. Applied Psychology, 71(4), 1171–1204. https://doi.org/10.1111/
apps.12241.

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Data-driven information for action 77

Pasmore, W., Winby, S., Mohrman, S. A., & Vanasse, R. (2019). Re- Vriens, D., & Achterbergh, J. (2011). Cybernetically sound organiza-
flections: sociotechnical systems design and organization change. tional structures I: de Sitter’s design theory. Kybernetes, 40(3),
Journal of Change Management, 19(2), 67–85. https://doi.org/10. 405–424. https://doi.org/10.1108/03684921111133656.
1080/14697017.2018.1553761. Wulff, K., & Finnestrand, H. (2022). It is like taking a ball for a walk:
Pava, C. H. P. (1983). Designing managerial and professional work for On boundary work in software development. AI & SOCIETY, 37,
high performance: a sociotechnical approach. National Produc- 711–724. https://doi.org/10.1007/s00146-021-01175-3.
tivity Review, 2(2), 126. Zuboff, S. (2019). The age of surveillance capitalism: The fight for
Pava, C. H. P. (1986). Redesigning sociotechnical systems design: con- a human future at the new frontier of power. London: Profile
cepts and methods for the 1990s. The Journal of Applied Behav- Books.
ioral Science, 22(3), 201–221.
Poland, B. D. (1995). Transcription quality as an aspect of rigor in qual-
itative research. Qualitative Inquiry, 1(3), 290–310. https://doi. Kristin Wulff has a master in
org/10.1177/107780049500100302. psychology (1996) and a master
Provost, F., & Fawcett, T. (2013). Data science and its relationship to in organization and management
big data and data-driven decision making. Big Data, 1(1), 51–59. (2014). She is now employed as
https://doi.org/10.1089/big.2013.1508. the Chief process and innovation
Riessman, C. K. (2008). Narrative methods for the human sciences. officer (CPIO) in Kantega and as
Los Angeles: SAGE. an Industrial PhD-candidate at the
Shani, A. B., & Coghlan, D. (2021). Action research in business Norwegian University of Science
and management: A reflective review. Action Research, 19(3), and Technology, Faculty of Eco-
518–541. https://doi.org/10.1177/1476750319852147. nomics and Management. She has
de Sitter, L. U., den Hertog, J. F., & Dankbaar, B. (1997). From com- worked within software develop-
plex organizations with simple jobs to simple organizations with ment in different roles since 1998,
complex jobs. Human Relations, 50(5), 497–534. https://doi.org/ the last 15 years in leadership posi-
10.1177/001872679705000503. tions.
Smith, R. G., & Eckroth, J. (2017). Building AI applications: yesterday,
today, and tomorrow. Ai Magazine, 38(1), 6–22. https://doi.org/
10.1609/aimag.v38i1.2709.
Teece, D. J., & Linden, G. (2017). Business models, value capture, and
the digital enterprise. Journal of Organization Design, 6(1), 1–14. Hanne Finnestrand is an associate
Trist, E. L. (1981). The evolution of sociotechnical systems as a con- professor of Organization develop-
ceptual framework and as an action research program. In A. H. ment at the Norwegian University
Van de Ven & W. F. Van de Joyce (Eds.), Perspectives on organi- of Science and Technology, Nor-
zation design & behavior (pp. 17–67). New York: John Wiley & way. She holds a PhD in Action
Sons. Research and Organization devel-
Trist, E. L., & Bamforth, K. W. (1951). Some social and psychologi- opment from the same university.
cal consequences of the longwall method of coal-getting: an ex- Finnestrand conducts action re-
amination of the psychological situation and defences of a work search in the fields of organization
group in relation to the social structure and technological content development, industrial relations,
of the work system. Human Relations, 4(1), 3–38. https://doi.org/ and socio-technical systems de-
10.1177/001872675100400101. sign. She is currently Editor-in-
Van Eijnatten, F. M. (1993). The Paradigm that changed the Work chief of the Springer Nature jour-
Place. Vol. 4. Assen: Van Gorcum. nal, Systemic Practice and Action
Vidgen, R., Shaw, S., & Grant, D. B. (2017). Management challenges Research.
in creating value from business analytics. European Journal of
Operational Research, 261(2), 626–639. https://doi.org/10.1016/
j.ejor.2017.02.023.

K
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

[email protected]