The proposed Directive on improving working conditions in platform work is gradually proceeding through the EU’s legislative institutions, with a key vote in the EP scheduled for 25 October. Beyond introducing protections in the case of algorithmic management, the most crucial innovation envisaged by the Directive is the stipulation of a conditional legal presumption of employee status for platform workers. What does national-level case law on the status of platform workers tell us about the prospects of this stipulation to improve these workers’ labour rights compared to the status quo?

Criteria triggering a presumption of employment

In essence, Article 4 as proposed stipulates the presumption of an employment relationship between platform and worker if two out of five indicators of “control” are found present. These criteria have not been picked at random, but chosen in the light of existing national-level jurisprudence on the status of platform workers, which to date entails more than 320 judgments and administrative decisions across Europe. These decisions evidence that, despite the diversity and complexity of the notion of employee in different national contexts, courts and administrative bodies in all of them have regularly attached decisive importance to one or more of these criteria when ruling on the status of platform workers.

“Effectively determining, or setting upper limits for the level of remuneration”

This is a key criterion, which has been referred to almost universally in case law, including at the highest level,[1]given that all national employee definitions contain elements of direction and authority exercised by a putative employer. For the vast majority of platforms concerned by such decisions – which were, in almost nine out of ten cases, either delivery or ride-hailing platforms – this criterion was fulfilled. By contrast, platform workers providing services in private households are often free to determine the fee to charge from the customers they serve, or to deviate from any suggested or minimum rate set by the platform. This freedom has regularly played a key role in the – so far scarce – rulings on these groups of workers, resulting in their qualification as self-employed in Denmark[2] and Sweden,[3] and as employees of the customer household in Norway.[4] The only court decision characterising such a platform as an employer – more specifically, a temporary work agency – is the Helpling ruling of the Amsterdam Appeals Court.

Developments in Spain – which has pioneered the strategy of supporting the classification of (some[5]) platform workers as employees via a legal presumption – indicate that platforms might rather change their policies so as to permit workers to determine their price than classify them as employees.[6]

“Requiring the person performing platform work to respect specific binding rules with regard to appearance, conduct towards the recipient of the service or performance of the work”

Such power to issue instructions is another element of direction and authority and has accordingly played a role in all judgments mentioned so far. Again, delivery and ride-hailing platforms have regularly been found to issue significant instructions, whereas workers performing cleaning, handyman services or errands for private households might receive instructions directly from the client – leading to the aforementioned controversial assessment in case law. For on-location microtasks such as mystery shopping, the scarce case law to date[7] indicates that instructions about concrete tasks are attributed to the platform even if its role is limited to distributing the requirements as formulated by the client.

Also in this regard, various platforms have adapted their business model in light of the “threat” of being classified as employers, by turning previously mandatory elements (e.g. concerning clothes and equipment, or the route to take) into voluntary guidelines. Such changes have regularly been invoked as justification for platforms to effectively ignore judgments that concerned an outdated version of their terms and conditions. Nonetheless, also decisions examining very recent versions of ride-hailing and delivery platforms[8] evidence that these are still based on an extensive set of instructions regarding required and prohibited conduct. Here, the outcome of the judicial assessment regularly depends on whether a court considers it relevant that the platform may have a vital and reasonable interest in guaranteeing the prescribed standards vis-à-vis the customer,[9] or finds that offering a highly standardised service is simply not legally possible when working with self-employed contractors. Note that the Council’s most recent compromise proposal on the proposed Directive suggests to exclude elements that are “necessary to protect the health and safety of the recipients of the service” from the assessment – which would apply to instructions aiming to ensure e.g. safe driving or food hygiene.

“Supervising the performance of work or verifying the quality of the results of the work including by electronic means”

This criterion would seem universally fulfilled by all platforms that have hitherto been subject to assessment of their potential employer status, as constant GPS tracking and rating systems – which in combination allow for a detailed insight into a worker’s performance – were applied even by those platforms which left the definition of tasks entirely to the client. Here, courts’ approaches could be divided into those which consider these structures for supervision relevant as such and those which require proof that the platform – or its algorithm – attaches any consequences to an indication of malperformance.

“Effectively restricting the freedom, including through sanctions, to organise one’s work, in particular the discretion to choose one’s working hours or periods of absence, to accept or to refuse tasks or to use subcontractors or substitutes”

This is another aspect in respect of which platform models have changed in recent years so as to appear less “employer-like”, as exemplified by the move from shift reservation to free login systems by platforms such as Deliveroo.[10] The scope of the concept of “effectively restricting… including through sanctions” will depend on whether it is considered to include models which reward a certain level or timing of activity on the platform – the effects of which have been illustrated in detail e.g. in German and Italian decisions.

Courts have been particularly divided with a view to substitution and/or subcontracting options. Notably the UK courts have seen them as sufficient in themselves to refute a claim of worker status, while other courts have considered theoretical replacement options non-decisive when they seemed impractical, or, as in one Dutch decision, even when proven relevant in practice but not suitable for “entrepreneurial” use.

“Effectively restricting the possibility to build a client base or to perform work for any third party”

Genuine exclusivity clauses have been considered incompatible with a claim of self-employment notably by some (but not all) French decisions. However, such clauses have become rare in the present-day platform economy. The now common partial restrictions (such as the prohibition to serve other customers while logged in on the platform’s app, or to exchange contact details with a customer found through the app) have been assessed controversially – from decisions finding them irrelevant because of their reasonable necessity for the platform’s business model[11] to those considering them equivalent to exclusivity if work for the platform requires a high level of activity in order to be profitable.[12]

All in all, it appears that, under a broad interpretation of the five criteria as proposed, virtually every platform model that has been subject to an assessment of employer status to date would be presumed to be an employer – including those specialising in household-related services or mystery shopping. Platforms offering a standardised service (notably ride-hailing and delivery platforms) would most likely be unable to escape the scope of the presumption even if a very strict interpretation was applied.

Legal presumptions as the strategy of choice for ensuring workers’ rights?

Legal presumptions which are designed to be difficult to rebut can have an immense effect on the classification of workers, as demonstrated by the famed Californian ABC test.[13] Also presumptions with an unrestricted rebuttal option can be of crucial relevance, notably in sectors characterised by a multitude of individualised and potentially poorly framed contractual relationships, for which employers may struggle to prove the essential facts and shy away from litigation over a single contract.

The platform economy, however, is characterised by companies imposing identical contractual conditions on a (very) large number of workers. The sheer scale of those conditions’ use means not only that platforms will regularly be more than ready to leave no remedy untried to avoid them from being declared unlawful, but also that there is regularly an abundance of evidence about the exact contents and circumstances of the contractual relationship. In the overwhelming majority of decisions on the platform economy, all major contractual features considered relevant by the decision-making bodies were common grounds between the parties of the dispute. In other words, the allocation of the burden of proof was irrelevant. Accordingly, already applicable legal presumptions of employee status have so far hardly played any role in the platform-related case law of the countries concerned (such as the Netherlands and Spain), and a Belgian decision of December 2021 became the first to demonstrate that a presumption, however broadly conceived, is effectively meaningless when rebuttable by a comparatively conservative notion of employee. And while the Belgian notion is rather unique in its emphasis on the will of the parties, elements that would presumably prevent typical cases of platform work from being characterised as continuous employment relationships can also be found e.g. in the employee concepts of Austria, Ireland and the UK, all of which attach much weight to the (personal) obligation to work.

Does this imply that the proposed Directive would be without effect? Certainly not. First, notably the introduction of rules on algorithmic management (Articles 6-8) is also relevant for those platform workers who end up being classified as self-employed, and various bodies involved in the legislative procedure[14] have already called for their extension to other (non-platform) workers. Second, also Article 4 might be expected to have, at the very least, a stimulating effect for legislators as well as courts at national level to reassess whether their traditional approaches to delimit the groups in need of legal protection are still in keeping with present-day labour market realities. In light of the principle of subsidiarity, common rules on a presumption of employment might be all the EU is allowed to do, considering that interfering with the actual notion of employee would have different effects in different national contexts. Given the wide variety of protective mechanisms in place for employees on the one hand and (subgroups of the) self-employed on the other across the EU, employee status may be either superfluous or even insufficient to guarantee platform workers the kinds of rights they are actually in need of.

Ultimately, the ball will again be in the court of the courts. Those have shown a remarkable readiness to explore new argumentative strategies in their platform-related case law in a number of countries,[15] and many of the most innovative rulings show an unusual degree of awareness of other countries’ case law.[16] These trends of “international cross-fertilisation” among judicial bodies might eventually fulfil an interesting function of complementing legislative harmonisation in the EU when labour markets are confronted with developments that constitute a challenge for all jurisdictions alike.



[1] Including the French Cassation Court (here and here), Italian Cassation Court, Spanish Supreme Court, Swiss Federal Court (hereand here), and UK Supreme Court. The Dutch Supreme Court is widely expected to follow suit in December this year, in line with the conclusions of the Advocate General.

[2] Decisions on Hilfr and Happy Helpers, overruling the platform’s self-qualification as an employer.

[3] Judgments on TaskRunner and Tiptapp, overruling decisions of the labour inspectorates.

[4] Judgments on Vaskerhvitt, overruling decisions of the labour inspectorates.

[5] The “Ley Rider” is restricted to delivery riders.

[6] This is the case for Glovo, the largest platform in the Spanish market, whereas UberEats is working with subcontractors and Deliveroo and Gorillas have left the market. The only notable platform hiring its workers as employees at present is Just Eat, which applies this policy also in other countries.

[7] Only lower-lever court rulings in Austria, but a Federal Labour Court decision in Germany and pending Cassation Court decision in France.

[8] In Belgium, Denmark, the Netherlands, or Switzerland.

[9] Cf. judgments in Belgium or Sweden (on Taskrunner and Tiptapp).

[10] Interestingly, the only two decisions expressly commenting on this move by Deliveroo both consider it irrelevant – finding, respectively, that the platform was an employer neither before nor after this change in Belgium, but both before and after it in the Netherlands.

[11] E.g. Swedish decisions on Taskrunner and Tiptapp.

[12] E.g. the French Cassation Court on Uber.

[13] Under this test, a rebuttal requires the cumulative proof of three elements, two of which (prongs B and C) are effectively alien to European concepts of employee status.

[14] The European Economic and Social Committee, the Committee of the Regions, and the European Parliament’s EMPL committee.

[15] Such as France, Germany, Italy, the Netherlands, Spain, Switzerland and the UK.

[16] Cf. examples in Belgium, Ireland, Italy, Spain, or Switzerland.

This page as PDF

Leave a Reply

Your email address will not be published. Required fields are marked *