Work(around) in progress

Have you already ignored some rules or recommendations today and done something your way instead? The question is whether this was a good idea or not and whether others could even learn from you. Workarounds like these are a fascinating field of research for information systems specialists at FAU. Their findings are now so advanced that they can improve processes on a long-term basis.

A small rectangular box in Prof. Dr. Sven Laumer’s car can help us to explain why a workaround, or a deviation from the standard process, can sometimes be useful. Laumer holds the Schöller Endowed Professorship for Information Systems (Digitalisation in Business and Society) and is on his way to work at FAU. He usually follows the instructions of this little box, which is his sat nav. Until today. That’s because the device’s recommendation to take a detour doesn’t actually result in the predicted time saving of seven minutes, as the professor later found out. Laumer ‘already had a hunch’ about this and decided to ignore the sat nav’s advice. He created his own route, his own personal workaround. ‘It doesn’t always make sense to follow standardised processes that are usually based on algorithms,’ he says, explaining his decision. However, the opposite sometimes also applies. It is quite possible that ignoring a well thought out and calculated recommendation can backfire.

Laumer knows what he’s talking about as his work involves dealing with workarounds. ‘There are IT systems in use everywhere that are supposed to offer support to individuals and industry with applications and services. However, in some cases they may never actually be used for the intended application scenarios. This is either because they are just not that useful or because they are used for a different purpose than the one they were developed for,’ he explains. The question is why this happens and which conclusions the field of information systems can draw from this.

A deviation from defined processes

Sven Weinzierl from the Chair of Digital Industrial Service Systems at FAU is conducting research on a similar topic. ‘Workarounds are a deviation from defined processes. They are deviating behaviour,’ he says. An example he cites involves the potential effects of a different member of a team ordering supplies for the office. He or she may work differently and may be able to secure better prices. However, the process may also prove less efficient. Another example is the question at which point choosing to fill in a form by hand instead of online (as intended) may in fact prove to be more productive. Or less complex. Personal reasons also play a role, such as laziness or scepticism towards new technologies.

‘In contrast to programmers or process coordinators in companies who seek to implement, optimise or standardise processes based on requirements and who may be puzzled when people bypass them, as researchers, we neither welcome nor criticise such workarounds,’ says Weinzierl. What Weinzierl and his colleague Laumer see and investigate is the optimisation value of a workaround. There are reasons behind each detour, and valuable insights can be gained by taking a closer look at them. This is either because they show the processes need to be improved, or because there are obviously more intelligent ways of achieving the same outcome than originally planned.

‘We want to understand how we can achieve high user satisfaction for work systems,’ says Laumer. What properties should a system have in terms of user-friendliness and content if it is to be widely accepted by users and seen as a welcome addition to existing technology? Laumer and his team have mainly used qualitative methods to try and answer this question. They have conducted a large number of interviews for over 40 case studies in order to collect data on existing processes. They then compared the work system (‘this is how I work’) with the information system (‘this is the support provided by the computer’). ‘The two systems often don’t operate hand in hand to the extent that was planned or expected,’ summarises Laumer. He quotes an example from a study published by his Chair around two years ago.

The main aim was to investigate how information resources stored in an enterprise content management system (ECM) at a bank and made available to the sales department were used. The researchers found that employees were not particularly keen on using the system to find basic information. Instead, employees had found a series of workarounds that led to an increased workload for the internal sales team. For example, staff at the bank were more likely to pick up the phone and ask one of their colleagues personally for up-to-date figures when approving loans, instead of getting the information from the database themselves. Laumer says this detour was not due to a general lack of interest in IT, but rather to the fact that the data required was stored in several different locations in the system, and was sometimes inconsistent. It was therefore quite logical and reasonable that staff sought some reassurance about which figures were up to date. In addition, the ECM system’s design was not really tailored to the needs of the sales department, as staff first had to find the information required by customer services and then compile it.

While specific examples like this can help to find an error in a certain system, the research and findings of Sven Weinzierl and his team are more fundamental. They have developed a method for automatically detecting workarounds in data. By using methods from deep learning, their method determines the standards in SAP-based processes, for example, and records deviations.

‘Deep learning, which is machine training of artificial neural networks using a large number of examples, has proven to be an excellent method for recording the variety of individual steps of a process and detecting patterns in the data,’ says Weinzierl. It is an effective means of analysing approval or order processes. However, before they can analyse this data, researchers first need to filter out some ‘background noise’. ‘This can be typing errors or even potential acts of sabotage,’ says the information systems expert. The detected workarounds are then evaluated. Workarounds with positive ratings are subsequently integrated into future standard processes. All workarounds with negative ratings are discarded.

Workarounds develop over time

In future, this technology will be developed in such a way as to allow evolving patterns to be detected. ‘Workarounds become institutionalised and develop over time without being consciously planned,’ says Weinzierl. This is comparable to a natural process, where it is also useful to know why it has developed the way it has.

General and practical findings can already be derived from the findings of this research. Prof. Laumer summarises them into three points. Firstly, poor usability repeatedly leads to workarounds – important issues for users do not receive sufficient support. Secondly, the information available via IT systems is often not precise enough to carry out tasks efficiently. Thirdly, systems often generate processes that later turn out to be unsuitable. That would be comparable with a sat nav that is not only difficult to operate but also provides more information about the weather than the traffic as well as information that is currently not relevant at all. Naturally, people who can think for themselves still get to where they want to go. However, it would be much easier if such workarounds were no longer necessary. Laumer and Weinzierl are working on it.

FAU research magazine friedrich

This article first appeared in our research magazine friedrich. You can order the print issue (only available in German) free of charge at presse@fau.de.

All articles[/button