Monitoring and Control: a Blessing for Employees?

Amazon operates a logistics centre in Winsen near Hamburg for the delivery of goods ordered by customers on the internet. Between 1,700 and 2,200 employees work there on an area of 64,000 square metres. On average, about 220,000 parcels a day are sent to customers from the logistics centre. Amazon offers its customers a delivery guarantee, so that there are only about four hours between the time the order is received and the time the package has to be handed over to the carrier at the latest.

The logistics centre is divided into different work areas where the employees' activities are subdivided into process paths such as unloading, receiving and storing goods as well as (on order) removing goods and packing them into packages and sorting these packages according to transport service providers. More than 90 per cent of the workers have to use hand-held scanners to document each of their work steps. The data collected in real time is used by Amazon to (1) control the current logistics processes and (2) the individual qualification of employees and (3) create objective assessment bases for individual feedback and personnel decisions.

After the data protection supervisory authority began monitoring the processes in 2017, it prohibited Amazon in October 2020 from continuously collecting current and minute-by-minute quantity and quality data of its employees and using it for the aforementioned purposes. The action brought against this by Amazon was successful at first instance before the Hanover Administrative Court (9 February 2023, 10 A 6199/20).

The court examined the permissibility of the data processing on the basis of § 26 BDSG, which essentially corresponds to Art. 6 para. 1 sentence 1 lit b DSGVO. According to this, the processing of personal data for purposes of the employment relationship is permissible if this is necessary for the fulfilment of the employment relationship.

Current and to-the-minute individual performance data of the employees would be needed for the control of the logistics processes in order to be able to react to performance fluctuations on the individual process paths by shifting employees between the work areas. 

On the basis of the performance data collected, Amazon is able to recognise in real time whether work on individual process paths is comparatively „fast“ or „slow” – in relation to the respective shift planning – and to react to this by reallocating staff in such a way that the flow of goods is approximately the same in all areas of the logistics centre and on all process paths, in order to be able to guarantee a uniform flow of goods through the entire logistics centre and compliance with all delivery guarantees given to Amazon's customers. 

In doing so, the different strengths of the employees would be perceived and taken into account depending on the assignment in the respective work areas, in comparison with other employees in the respective work area and in comparison with the overall individual performance. Only in this way could Amazon compensate for any deficits (overload or underload) on an ad hoc basis by redeploying particularly qualified employees who would be seconded from another workplace at the time. Using the current data, it could also locate all employees when needed without having to spend time searching for them in the logistics centre. In this context, the court held that only real-time monitoring of the activities of all employees would enable such a control, whereas a mere control of the flow of goods instead of the employees would not be sufficient. 

According to the court, a comparable control without the possibility of real-time monitoring and taking ad hoc measures was not possible. Taking into account only average values of the employees' performance, based on previous random samples, would be associated with significant losses of efficiency. This is because the aggregated values (necessarily) do not provide information on the shift-specific and current performance capacity of an employee. However, the determination of the current and individual performance capacity of the employees on the different process paths is crucial for the control of the logistics centre, because the control consists precisely in the reaction to the deviation of the actual situation from the planned situation based on average values before the start of the shift by shifting employees along the entire processing chain. The necessity of the data processing would thereby be explicitly measured against Amazon's operating procedures, according to the Administrative Court.

In the court's opinion, the permanent qualification of employees also requires continuous monitoring of their work performance. As a matter of principle, Amazon deploys employees in all work areas on a rotating basis. In order to be able to detect the need for further training immediately, this need had to be determined on the basis of the performance data collected in real time. Based on the data, it is easy for Amazon to see whether an employee is already capable of working on a new process path at an appropriate speed. If Amazon did not collect this data, it would be significantly limited in the distribution of employees across the various process paths. It would (presumably) be dependent on deploying employees on only a few process paths or even training them in test operations first. In any case, Amazon would then be prevented from reacting as flexibly as it currently does to the different needs in the various work areas and on the various process paths, which would result in considerable efficiency losses, so that Amazon could no longer operate the logistics centre as before.

Finally, the court accepts as necessary the continuous recording and evaluation of the performance data of Amazon's employees for the feedback processes and for the preparation of personnel decisions. In this context, staff members receive recognition feedback (highest-performing ten per cent) or development feedback (lowest-performing five per cent) every two weeks, training feedback within the first ten weeks on new process paths and, in any other case, in addition, quarterly feedback for all other staff members.

Continuous feedback discussions in the employee relationship are part of the established principles of good employee management. Due to the operational circumstances, Amazon is dependent on personal performance data in order to be able to give fact-based feedback at all. Feedback proposals based on merely aggregated averages instead of precise measurements would be less meaningful and thus worse, would reduce the added value of the feedback and – according to the court – cannot be intended by data protection law.

The supervisory authority complains that the continuous data collection leads to a permanent pressure on employees to adapt and perform, which is unreasonable. The court disagrees: The pressure would not be less if the data collection were limited to only half of the working hours, for example. Without the data processing systems, it would not be possible for Amazon to intervene in the process flows in a fine-tuned manner. At various points along the processing chain, goods backlogs or goods shortages would occur, jeopardising the smooth processing of goods at Amazon. This would not only put the employees concerned under considerable stress. Rather, it would have considerable consequences for company peace as a whole, since the data processing systems ensure that the work that arises in the logistics centre can be done „reasonably calmly“ – according to the works council chairpersons. In the court's view, the operation of IT assistance systems such as those at Amazon is therefore associated with advantages, as friction in the work processes is reduced, an efficient distribution of work among the employees is guaranteed and the overall workload for the employees is reduced. A simple conclusion from the „permanent“ operation of an assistance system on the one hand to an excessive monitoring burden on employees affected by it on the other hand is therefore expressly described by the court as „under-complex“.

In the end, Amazon has put itself in a position in which the data processing carried out is arguably necessary in the context of efficiency: Precisely because Amazon makes a certain delivery promise to customers, close-grained data must be collected and used, especially for process optimization. The rest are follow-up questions.

It remains to be seen whether such an approach is actually possible and can thus also be implemented by other market participants. The court has allowed the appeal because of the special importance of the matter. The supervisory authority has already filed an appeal.