The routine
aspect of work has emerged as a key construct in many subfields of management
and organization studies. Fields such as strategic management (Felin &
Foss, 2005), international management (Haleblian et al., 2009) and organization
theory (Salvato & Rerup, 2011) have gained new insights by utilizing the
routine construct. Despite these advances in multiple fields, the literature on
routines is still riddled with ambiguities (Abell, Felin & Foss, 2008; Becker,
2004; Becker, 2008; Felin & Foss, 2009). Arguably, the ambiguity between
action and potential action, the ambiguity between individual and collective
level, and the ambiguity between stability and change are the most significant
hindrances that deter the theoretical development of the routine concept
(Becker, 2008). Recently, studying the micro-level origins of routines has been
proposed as a robust candidate for answering the obstacles related to the
theoretical development of the routine concept (Abell et. al., 2008; Felin
& Foss, 2005; Felin & Foss, 2009; Gavetti, 2005).

The need for studies on microfoundations of routines stems from the idea
that many subfields of organization theory have began at some aggregate
analytical level N at time t (Nt) in their early stages of development and thus
implicitly assumed that micro-level (N-1t-1) phenomena have relatively uniform
effects on aggregate level phenomena, for example uniformity among workforces in
population ecology and culturally doped employees in institutional theory
(Felin, Foss, Heimeriks & Madsen, 2012, 6–7). Consequently, by unpacking
routines in microfoundational terms we will gain an understanding of routines’
different constituent components.

Possibly the most comprehensive attempt to map the constituent
components of routines has been Journal
of Management Studies’ special issue ‘Micro-Origins
of Organizational Routines and Capabilities’ (2012). Based on several
recent theoretical and empirical studies that have devoted explicit attention
to the micro-level origins of routines (Becker and Lazaric, 2003; Becker,
Lazaric, Nelson & Winter, 2005; D’Addiero, 2009; Gavetti, 2005; Salvato,
Sciascia & Alberti, 2009; Rerup & Feldman, 2011; Feldman & Pentland,
2003, Pentland & Feldman 2008; Teece, 2007) the editors of the special
issue propose three constituent components of routines as a starting point for
the microfoundational research agenda: 1) individuals, 2) processes and
interactions, and 3) structure and design. Further, the editors conclude that the
research agenda should focus on interactions between these three constituent
components.

The
purpose of our study is to construct a model of organizational routines’ microfoundational dynamics. This
model combines the constituent components  – individuals, processes and interactions,
structure and design – of organizational routines (Felin et al., 2012) with the
major dynamics – variation and selective retention – that shape organizational
routines (Pentland,
Feldman, Becker & Liu 2012). As a part of the model, we examine the role of automatic human
behavior in the workers who carry out their parts in organizational routine. As
a result of this study, we discuss the model of organizational routines’ microfoundational dynamics in
relation to the most significant ambiguities: individual versus collective
level, action versus potential action, and stability versus change.

Our findings are based on a
combination of quantitative and qualitative approaches on the data collected
from the top professional ice hockey league in Finland. This league has recently
invested in such digital technologies as a smart puck and wearable transmitters,
which have significantly transformed the way in which game statistics can be
generated. From a theoretical perspective these advanced game statistics
represent structures, which will contain less variance than traditional
statistics or rankings. Alongside with quantitative data, namely the advanced
statistics, we interview and observe professional ice hockey coaches and
players to make sense of how they experience the developmental steps in the
digitalization and the resultant changes in their work routines. Consequently,
we hypothesize that this more detailed information will change the work of
coaches and players. Team sport provides us a particularly transparent research
context to observe these probable changes in routinized work.    

Our
study contributes the debate between studies in which routines are seen as
actions (eg. Pentland, Haerem & Hillison, 2010; Salvato & Rerup, 2011)
and studies in which routines are seen as potential actions (eg. Birnholtz,
Cohen & Hoch, 2007; Hodgson, 2008; Knudsen, 2008) by illustrating how these
two previously incompatible viewpoints can be combined in one model, which
attempts to estimate the path dependency inside the routine with probabilities.
Specifically, this idea – originally presented by Pentland et. al. (2012) – is
applied to incorporate the path dependency between routine performances by demonstrating
how different structures retain different amounts of the endogenous variance in
routine. Second, we contribute the stability versus change discussion (Becker,
2004; Cohen, Burkhart, Dosi, Egidi, Marengo, Warglien & Winter 1996.) by
examining how both human and non-human actors affect variation and selective
retention in the iterations of the routine (cf. Pentland et. al., 2012; Turner
& Fern, 2012).