In this chapter
Employees who have developed workarounds may well have no incentive to disclose them, especially if the ethos of the organisation is to see the use of workarounds as being an example of resistance to change. In this chapter the use of both qualitative and quantitative approaches to the discovery of the existence, scale, value and potential risk are considered. Applications that log process steps to highlight potential workarounds in use have limitations when it comes to workarounds for procedures.
Can you count workarounds?
How many workarounds are there to discover? Almost certainly more than you might imagine. In the many case studies that have been undertaken usually only a small number of employees are interviewed, a very small percentage of the total number of employees. The choice of employees to interview is made by the organisation, not the research team, so the end result is in effect a somewhat random sample. Yet as the interviews proceed most, if not all, of the interviewees have a story to tell. There is a paradox in that until there is a reasonably clear understanding of the scale of workarounds it is not possible to decide on the balance between the two approaches. It could be that ten workarounds are discovered out of perhaps hundreds that might be in active use. There is a parallel situation in search when seeking a complete recall of all relevant documents; there is no way of knowing how many there actually are to assess whether the search has been successful.
However it is not just a question of how many. In your organisation there might be just a few but they are being used in processes which could have a serious impact on performance and reputation if the workaround is inappropriate. The number of workarounds is also subject to rapid change as new systems are introduced and older systems are upgraded. Should the attention only be to making the best of the investment in a new application and overlook workarounds in legacy applications that might be generating significant technical debt and corporate risk?
The discovery process also has to be linked to a remedial process. If the decision is taken that a particular workaround needs to be eliminated then what is going to take its place? Nothing will annoy an employee more than being ‘found out’, criticised by their manager and told that using the designed system is mandatory.
Making the invisible visible
There are many challenges to overcome when embarking on a project to discover workarounds within an organisation. For a start there are two quite distinct discovery methodologies, each of which has value. The first is to use data logging to create a quantitative assessment of the incidence of workarounds at an individual process level. The second is to use an ethnographic approach to understand the factors affecting the development and adoption of workarounds at an employee level. In practice it needs a blend of both but determining the balance between the approaches is difficult. There are no reliable independent surveys of the extent of workarounds in organisations but as discussed in Chapter 7 there are surveys of shadow IT adoption. These surveys indicate that the use of shadow IT is widespread, and after all shadow IT is arguably a workaround.
The initial challenge is to decide which applications and which processes to explore for workarounds given the range of applications in use in an organisation. A report released in 2021 by Productiv, a provider of applications to identify shadow IT, indicates that within their SaaS (Software As A Service) library large enterprises average 364 applications, while small businesses average a portfolio of 242 applications. Productiv analysed 107 categories of applications in its survey and found that organisations typically had 17 categories with five or more tools in critical categories like project management, sharing and storage, and messaging. It is not surprising that enterprise search logs show a high incidence of employees looking for applications by name or by function. Concur is the brand name of an SAP expenses management application and often occurs well up the list of most searched terms on an enterprise search log. It may only be used monthly so employees find it difficult to remember how to locate a very important (to them!) application which is used relatively infrequently.
To identify a workaround in a process it is first necessary to have a well documented description of the process and the potential variations in how it can be used. There has been a substantial amount of research into business maturity models (Tarhan 2016) but in practice a wide range of models are in use. The Microsoft Model for Microsoft 365 Business Process Maturity is an example. The question is at what level of maturity can a data logging application provide a grounded basis for deciding that a workaround is being used?
Despite the number of processes supported by IT applications there are also a substantial number of much more flexible and poorly documented procedures. An example might be the compilation of a user manual for a high technology product which will go through many iterations and involved multiple review processes, probably managed by circulation of each version attached to an email.
A paper by Beverungen et al (2021) sets out seven paradoxes of business process management in what they describe as a hyperconnected world, mentioning (for example) the role played by smart personal devices.
Design science project framework
A number of studies into workarounds have adopted what is referred to as design science research model to frame the investigation and the analysis of the outcomes. From a workaround perspective the benefit of using this model is that it starts with identifying requirements and then moves on to designing the solution. Although this approach has been developed and tested within academic research the steps could also be of value to an organisation in setting up a research project to assess the prevalence of workarounds.
Pello (2018) provides an introduction to design science research which is built around a seven-step process.
To quote from the author’s commentary
“First, carry out end-user researchto gain insights and discover the active and latent needs and values of the users, and understand the factors of behaviour (what do people think, why they do what they do or do not do what they are supposed to do, what are their attitudes towards the problem, their belief systems; and cultural, political, legislative and social context; etc.).
Second, define clear objectives and restrictionsbased on the findings (does the solution need to be a new physical object, label; or an intangible service or a process according to which something is made easier; etc.).
Third, using different techniques (like brainstorming, experience sketching, feature trees, etc.) gather different ideasfor the solution.
Fourth, filter out the viable and feasible ideasfor testing (evaluate the ideas).
Fifth, test the chosen ideas with the end-users to find out the best solution (do the end-users understand the solution or not; can they use it without extra instructions; etc.).
Sixth, iterate by reviewing, refining and retestingthe solution in order to get to the best possible solution that can be generalised.
Seventh, compare the solution with theories, develop on the existing theories, generalise the outcomeand share the knowledge with appropriate audiences (people, companies and policy makers).”
Although organisations may undertake employee satisfaction surveys, the lack of expertise in writing research surveys (as distinct from employee satisfaction/opinion surveys) is often very noticeable. There is usually even less experience with interviews and the associated elements of ethnographic research. Ethnography is a social studies research methodology based on observing the behaviour of the participants in a given social situation and also understanding the group members’ own interpretation of this behaviour. The methodology dates back to the mid-1740s.
Madden (2017) provides a good introduction to ethnographic research.
Although ethnographic research can be of great value in understanding why certain workarounds have been introduced, the research process needs to be developed, managed and analysed with considerable care to ensure that the right blend of techniques is adopted and adapted as evidence is collected. It is not just a question of circulating a survey or undertaking some interviews without a clear set of objectives and having staff with the skills to conduct the interviews and analyse the outcomes.
Ethnographers use a range of methods depending on the situation or need to gain different slices of understanding a target group or situation of interest. These typically include
- Semi-structured or in-depth interviews
- Asking employees to demonstrate and explain the approaches they are describing
- Asking exploratory questions in the process of observing employees go about their usual activities to gain context on their actions
Xerox PARC has played an important role in the development of IT but its contribution to industrial ethnography is far less well appreciated. Xerox PARC was a pioneer in hiring social scientists into corporate R&D and integrating them among its technological staff. In effect it sparked an interest in what might be regarded as industrial ethnography as distinct from social ethnography.
One of the pioneers of ethnography at Xerox PARC was Richard Harper. In 1998 he wrote ‘Inside the IMF’ (Harper 1998) which remains the only comprehensive ethnographic study of information management in a single organisation, in this case the International Monetary Fund. The book is subtitled ‘An ethnography of documents, technology and organisational action’ and starts with an introduction to the evolution of ethnography and in particular the value of organisational ethnography. There is a passing reference in the book to workarounds but in the mid-1990s, when the research for this book was undertaken, document management technology was in its infancy.
Ducheneaut et al (2010) provide an introduction to ethnographic research in what they term ‘virtual worlds’. The final section of this overview looks specifically at using a digital ethnography tool kit to solve business problems, though without specific reference to the discovery of workarounds.
Although not specifically concerning workaround discovery Gupper and Mörike (2022) consider the role of internal social media channels in supporting ethnographic research.
“While digital communication platforms enable researchers to communicate with research participants across large distances, or observe digitally mediated interactions at play, our results highlight the limits researchers face when employing such platforms in their research. Hybrid settings, where communication flows are both in-situ and digitally mediated, further increase= the complexities. An understanding and reflection of these limits should thus be an integrative part of any ethnographic fieldwork makes use of digital communication platforms”
The authors present a three-level model of digital visibility in ethnographic field work, namely Invisible, Uncertain and Visible and conclude with four questions that should be considered by any person developing a digitally supported ethnographic research project
- What aspects relevant for my research question can remain hidden if I choose to conduct only digitally mediated research?
- What connotation(s) do(es) the digital communication platform(s) carry in the context in which I conduct research, and how will this meaning ascribed to the platform influence the insights I can gain there?
- What forms of communication do(es) the digital communication platform(s) I intend to use for my research enable, [and] what forms are not supported?
- In which physical context can I perceive which forms of communication in a hybrid setting, and what might remain either uncertain or invisible to me?
Over the last decade computational ethnography has emerged to offer a wider range of quantitative techniques through data logging and process tracking and also enhance the value and veracity of diary studies based on randomly timed requests to an employee to complete a survey response, or to have the response request triggered by a specific action.
Van der Schaft–Bartis (2013) made an important contribution to the use of ethnographic techniques in her 2013 thesis. Section 4 of the thesis considers the options available to researchers when investigating organisational processes with comments on the role and challenges of each.
Shaft-Bartis offers a very important perspective on research methodology
“An important factor is that, although I managed to develop a good relationship with the research participants, the collected data was possibly influenced by their interpretation of (1) the term “workaround” and (2) my research and its consequences. They might have forgotten, or decided to rate unimportant, unnecessary – or risky – to share certain tricks with me. This might be in the background of having found a bit less individual solutions than I expected – both during the interviews and the observation. Although the method of observation somewhat counterbalances the possible congruence between their actions and the story told, but due to technical details I sometimes had to ask questions to complement the observation – this made the observation less neutral and less “invisible”. Therefore, it has to be highlighted that the collected data is very much defined by the explanations of the users. This window for biases brings some weakness to the reliability of the collected data.”
The author goes on to note;
“It is important to mention that I entered both companies through connections to the Managing Directors. I have to assume that as a consequence, my person, my presence and my research was also connected to the top management. This might result in the participants being less open with me – with or without intention. Naturally they were not able to see the consequences of showing me a practice what might be forbidden. As a further result, the top management perspective is strongly present in the thesis.”
These caveats are quoted in full as they need to be taken into account in any ethnographic research.
This raises the important issue of whether to use internal staff resources to undertake an ethnographic study or to out-source the project.
Some recent research papers by Mörike (2013) provide an excellent introduction to the value and challenges of ethnographic research. In her initial contribution to the research literature presented the concept of working misunderstandings. A research project undertaken in India is described which uncovered some differences in the ways in which different teams worked on a project. Although the term ‘workarounds’ is not specifically used, the paper is a valuable introduction into the use of ethnographic research, especially where there is an emphasis on direct observation.
More recently two case studies (Morike 2022) are reported, one within a small engineering company and the second in a clinical healthcare setting. Both papers provide a wealth of detail into the processes, benefits and challenges of ethnographic research methods.
Alfredo (2022) described in detail the training that is required to undertake direct observation of processes in use. The focus is on tracking surgical processes in hospitals but the advice given on the importance of training observers has a much wider applicability, going right to qualitative edge of the quantitative-qualitative spectrum.
Over the last decade the development of business process modelling and process mining has been very rapid in terms of both capability and availability. Van De Aalst (2013) provides a good introduction to the technology of business process mining prior to the recent adoption of machine learning technologies. Van De Aalst is also the author of a book on process mining (2016). Process mining records (usually on a time line) the duration of each step of a process. Task mining records the interactions between the employee and the desktop, tracking key strokes and migration between applications. The end result of both is a substantial database of log data which is going to take both time and a detailed knowledge of each process to identify potential workarounds.
The initial work on this approach was undertaken by Outmazgin (2013). This paper is important to consider as the authors categorise workarounds into six categories of which only four can be detected by data logging.
The two cases where detection was regarded as not feasible were
Type B – Selecting an entity instance that fits a preferable path
This type of workaround relates to situations where a “legitimate” process execution is performed, but the entity instance that is used does not represent the actual one. Rather, it is chosen in order to comply with the transition conditions of the process.
Type F – Separation of the actual process from the reported one
In this workaround type, at a certain stage the process participants continue the process manually, possibly until the process is completed. At a separate point in time, the actions that were performed (or should have been performed) are reported in an orderly manner. This is done in a post-hoc manner, only for the purpose of documentation and reporting.
The authors conclude their paper by commenting
“Developing an understanding of the workarounds that take place and particularly of the reasons that drive them would be valuable in improvement efforts. Corrective actions can include redesigning the processes, improving the data flow, the permission and control mechanisms, role definitions, and also training and disciplinary actions. This is expected to lead to improved performance as well as compliance. Future research will aim at investigating the reasons for workarounds, and establish relationships between process properties, such as bottlenecks and number of participants, and the frequency of workarounds.”
A later paper (Outmazgin 2016) reflects on this research project. In the conclusion the comment is made
“We note that considering our notion of work-arounds, the detection might include both false positives, cases that are falsely indicated as work-arounds, and false negatives, actual work-arounds that are not detected. Specifically, we define work-arounds not just as incompliant behavior, but as one that involves intentional defiance of known procedures. Clearly, we have no means for assessing user intention from event logs. To this end, we rely on the list of work-around types, which was obtained through interviews where users indicated what they perceive as work-arounds. It might be that the resulting patterns also include incompliant behavior performed for different reasons.”
That is a very honest assessment but it inevitably raises issues for an organisation. False indications may result in employees being challenged to justify the approach they are taking when in fact they are working compliantly. False negatives could result in high-risk workarounds not being detected and addressed.
Quantitative research using data logging and process mining might well give a sense of scale of workarounds but may not even identify the employees undertaking the workaround. This is especially the case where there is a use of shadow IT to undertake a process (the ever-useful Excel file) that does not show up on the process mining dashboard.
Deep learning approaches
Over the last decade there has been considerable progress in using AI/machine learning approaches, often embedding the outcomes within a Design Science Research (DSR) framework. Two good introductory papers to the logging methodology come from Weinzierl et al (2020 and 2022).
An important forum for the presentation of research into business process modelling is the annual Business Process Management conferences which take place in Europe and started in 2003.
As an illustration of the scope of the conferences the topic sessions in the 2022 conference were
- Task Mining
- Design Methods
- Process Mining
- Process Mining Practice
Typically there are around 30 papers presented at these conferences as well as tutorial workshops.
Small and medium-sized organisations
Undertaking workaround discovery in small and medium-sized organisations is the subject of a paper by Wijnhoven (2023). In these smaller organisations processes may be more ad hoc and less well documented. This paper provides a description of workarounds discovered in the course of a research project at an engineering company with 170 employees. The conclusions of the authors are
- Process mining in smaller organisations can be particularly challenging because of the informal nature of these organisations, which leads to a less complete de jure process model and under-developed process-aware system semantics.
- It can be difficult to classify non-compliance cases as workarounds. Fraud and obstruction may remain hidden.
- Evaluating different categories of workarounds can be beneficial for determining priorities or management actions related to workarounds. However, the role of process mining in this context is limited and human insights (e.g. interpretations) in the broader context of the work system processes are necessary.
Most of the research into implementing business process management applications to detect workarounds has been in enterprise information systems. Workaround detection is of great importance to the use of Electronic Health Record systems in hospitals and primary care facilities. An important contribution in this sector has been made by Beereport (2021) primarily based on her PhD thesis (Beereport 2021). This thesis is based on a comprehensive literature review of over 250 research papers together with empirical investigations at a major hospital in the Netherlands.
A subsequent paper (Van Der Wall et al 2022) presents the development and utility of SWORD, an acronym for a semi-automated WORkaround Detection (SWORD) framework. Of particular value is a table of 22 log patterns which might indicate the use of a workaround.
This research has been partially funded by the WorkAround Mining Lab of the University of Utrecht through NWO Open Technology Project “WorkAround Mining (WAM!): Mining the emergence, evolution, and diffusion of workarounds in health information systems” (Project Number 18490). The objective of this Laboratory is to investigate the emergence, evolution, and diffusion of workarounds in organisations. The projects adopt different research methods, such as interviews, observations, and process mining.
Process vs information
There is a fundamental problem with logging-based applications and that is that the focus is on time taken, and to some extent the paths through related processes, but there is no tracking of the content itself. As a result information workarounds cannot be detected and (as discussed in Chapter 9) these potentially carry a much higher corporate risk. It should also be appreciated that employees with a neurodivergent condition may have time-blindness as a result. They may not be able to judge the passing of time, work to a very closely defined time-line for a process step and may also need a longer time to work through the options for a process step.
This is a particular problem in clinical Electronic Health Record applications where an error in the notes on a patient could have serious consequences. In EHR logs it is usually possible to detect free text outside of a text box but with no ability to check on the accuracy of the text. This also gets into data privacy issues where access to patient records is very tightly controlled.
Integrating qualitative and quantitative research
The integration of qualitative and quantitative research is often referred to as a ‘mixed methods’ approach. There is a substantial literature on this subject, a number of books, and a research journal, Journal of Mixed Methods Research.
The bottom line
The balance between quantitative (data logging) and qualitative (surveys and interviews) methodologies is very difficult to determine at the start of a discovery project and may need to be modified in the course of the project. Using process mining for small and medium-sized organisations runs into many challenges as the processes are often not well defined. In Chapter 5 research into the use of workarounds to enterprise systems is presented
Abb, L., & Rehse, J-R. (2022). A reference data model for process-related user interaction logs. 20th International Conference on Business Process Management (BPM 2022) September 11–16. Lecture Notes in Computer Science, 13420, 57-74 https://arxiv.org/abs/2207.12054
Alfred, M., Del Gaizo, J., Kanji, F., Lawton, S., Caron, A., Nemeth, L.S., Alekseyenko, A.V., Shouhed, D., Savage, S., Anger, J.T., Catchpole, K., & Cohen, T. (2021). A better way: Training for direct observations in healthcare. BMJ Quality & Safety, 31(10) https://qualitysafety.bmj.com/content/31/10/744
Antunes, P., Pino, J.A., Tate, M., & Barros, A. (2020). Eliciting process knowledge through process stories. Information Systems Frontiers, 22, 1179–1201. https://link.springer.com/article/10.1007/s10796-019-09922-0
Bade, F.M., Vollenberg, C., Koch, Jannis, Koch, Julian, & Coners, A. (2022). The dark side of process mining. How identifiable are users despite technologically anonymized data? A case study from the health sector. 20th International Conference on Business Process Management (BPM 2022) September 11–16, 2022. Lecture Notes in Computer Science, 13420, 218-233. https://link.springer.com/chapter/10.1007/978-3-031-16103-2_16
Beerepoot, I.M. (2021). Workaround: The path from detection to improvement. [PhD thesis, Utrecht University]. https://dspace.library.uu.nl/handle/1874/416626
Beverungen, D et al (2021). Seven paradoxes of business process management in a HyperConnected world. Bus Inf Syst Eng, 63(2):145–156. https://doi.org/10.1007/s12599-020-00646-z
Bellan, P., Dragoni, P., & Ghidin, C. (2021). Process extraction from text – state of the art and challenges for the future. http://ceur-ws.org/Vol-2776/paper-3.pdf
Calvanese, D., Lukumbuzya, S., Montali, M., & Simkus, M. (2021). Process mining with common sense. https://www.diva-portal.org/smash/get/diva2:1595958/FULLTEXT01.pdf
Dubinsky, Y., Soffer, P. & Hadar, I. (2023). Detecting cross-case associations in an event log: toward a pattern-based detection. Softw Syst Model. https://doi.org/10.1007/s10270-023-01100-w
Ducheneaut, N., Yee, N., & Bellotti, V. (2010). The best of both (virtual) worlds: using ethnography and computational tools to study online behavior. Ethnographic Praxis in Industry Conference, 1, 136-148. https://doi.org/10.1111/j.1559-8918.2010.00013.x
Gupper, T & Morike, F. (2022). Visible – uncertain – invisible: Reflections on team communication flows in digitally mediated ethnographic fieldwork. NordiCHI ’22: Nordic Human-Computer Interaction Conference October 2022, 59, 1-9. https://doi.org/10.1145/3546155.3546674
Harper, R.H.R. (1998). Inside the IMF. An ethnography of documents, technology and organisational action. Academic Press. ISBN 0-12-325840-5
Madden, R. (2017). Being ethnographic. A guide to the theory and practice of ethnography. 2nd edition. Sage Publications Ltd.
Marchavilas, G. (2020). In search of an appropriate notation to model business process workarounds: The effect of BPMN elements on understandability and user acceptance. Thesis. Utrecht University Graduate School of Natural Sciences.
Martínez-Rojas, A., Jiménez-Ramírez, A., Enríquez, J.G., & Reijers, H.A. (2022). Analyzing variable human actions for robotic process automation. In: C. Di Ciccio, R. Dijkman, A. del Río Ortega, & S. Rinderle-Ma (eds). Business Process Management. BPM 2022. Lecture Notes in Computer Science, 13420. Springer https://link.springer.com/chapter/10.1007/978-3-031-16103-2_8
Mörike F. (2013). Working misunderstandings and notions of collaboration. https://journals.openedition.org/civilisations/4081
Mörike, F. (2022). Inverted hierarchies on the shop floor: The organisational layer of workarounds for collaboration in the metal industry. Computer Supported Cooperative Work (CSCW), 31, 111-147. https://link.springer.com/article/10.1007/s10606-021-09415-2 https://doi.org/10.1007/s10606-021-09415-2
Mörike, F., Spiehl, H. L., & Feufel, M. A. (2022). Workarounds in the shadow system: An ethnographic study of requirements for documentation and cooperation in a clinical advisory center. Human Factors. https://doi.org/10.1177/00187208221087013
Outmazgin, N., & Soffer, P. (2013). Business process workarounds: What can and cannot be detected by process mining. In Enterprise, Business-Process and Information Systems Modeling, 48-62. https://link.springer.com/chapter/10.1007/978-3-642-38484-4_5
Outmazgin, N., & Soffer, P. (2016). A process mining-based analysis of business process work-arounds. Software & Systems Modeling, 15,309-323. https://link.springer.com/article/10.1007/s10270-014-0420-6
Pello, R. (2018). Design science research – a short summary. https://medium.com/@pello/design-science-research-a-summary-bb538a40f669
Pelz-Sharpe, A., & Loubani, A. (2021). Process and task mining: vendor landscape 2021-22 Deep Analysis https://www.deep-analysis.net/
Tarhan, A., Turetken, O., & Reijers, H. A. (2016). Business process maturity models: a systematic literature review. Information and Software Technology, 75, 122-134. https://doi.org/10.1016/j.infsof.2016.01.010
Van der Aalst, W.M.P. (2013). Business Process Management: A comprehensive survey. ISRN Software Engineering. Article ID 507984. http://dx.doi.org/10.1155/2013/507984
Van Der Aalst, W.M.P. (2016). Process mining – data science in action. ISBN: 978-3-662-49851-4
Van der Schaft–Bartis, E. (2013). Means of interpretive flexibility: User workarounds next to information systems. PhD thesis. Budapest Corvinus University.
Van der Wall, W., Beerepoot I., Van de Weerd, I and Reijers, H.A (2022). The SWORD is Mightier Than the Interview: A Framework for Semi-automatic WORkaround Detection. BPM 2022, LNCS 13420, 91–106, 2022
Weber, S. (2010). Design science research: paradigm or approach. AMCIS 2010 Proceedings, 214. https://core.ac.uk/download/pdf/301344775.pdf
Weinzierl, S., Wolf, V., Pauli, T., Beverungen, D., & Matzner, M. (2020). Detecting workarounds in business processes: a deep learning method for analyzing event logs. In ECIS (European Conference on Information Systems) 2020 Proceedings. https://aisel.aisnet.org/ecis2020_rp/67/
Weinzierl, S., Wolf, V., Pauli, T., Beverungen, D., & Matzner, M. (2022). Detecting temporal workarounds in business processes: A deep-learning-based method for analysing event log data. Journal of Business Analytics, 5,(1), 76-100. https://www.tandfonline.com/doi/full/10.1080/2573234X.2021.1978337
Wijnhoven, F., Hoffman P., Bemthuis, R. and Boksebeld, J. (2023). Using process mining for workarounds analysis in context: Learning from a small and medium-sized company case. International Journal of Information Management Data Insights 3 , 100163.