Skills Development 4.0: What can we learn from workers in the online platform economy?

by Konstantinos Pouliakas, Expert in Skills and Labour Markets, Cedefop*

Debates about the platform economy have mainly focused on issues of gig workers’ lack of social insurance and dubious employment classification. Both policy and academic literature have been relatively silent in their analysis of the skills development and matching processes in the online platform economy. Many relevant questions about the skills demanded in online platform work have been left unanswered: Is the platform economy a digital sweatshop, devoid of a typical organizational context, in which workers are requested to carry out mundane tasks, entailing a-race-to-the-bottom in terms of skill needs? Or is it an environment conducive to skills development because of global talent competition, lifting of entry barriers to work and opportunities to carry out high-skill projects?

To provide insights on the skills demanded and learning opportunities in the online platform economy, the European Centre for the Development of Vocational Training (Cedefop), together with researchers from the Oxford Internet Institute and Copenhagen Business School, initiated the so-called “CrowdLearn” project in 2017. The project involved a mixed-method research approach (qualitative and quantitative survey phases) that drew information from five main online labour platforms — UpWork, PeoplePerHour, Fiverr, Twago, and Amazon Mechanical Turk — and focused on EU-based workers.[1] At the time the project began, the EU scope was a novelty because most literature had focused on crowdworkers from Anglo-Saxon countries or Asia.

The research focused on two groups of crowdworkers — macroworkers (mostly online freelancers, engaging in relatively high-skill tasks such as ICT, creative design, writing/translation and sales or marketing) and micro-workers (engaging in relatively repetitive tasks, such as survey implementation, web information retrieval, image tagging etc.).[2]

Specific questions that the project posed were: what skills online platform workers need and develop? Why do they engage in online platform work and how does this affect their incentives for skills accumulation? How does skill development and matching take place in the online platform economy? With whom do online platform workers engage in skill formation? What can vocational education and training (VET) policy do?

In terms of the characteristics of the Crowdlearn sample of crowdworkers, it is found that platform workers in general possess high levels of human capital; they consist of mainly high-educated individuals (many with a postgraduate degree). When it comes to their main employment status, 45% of the surveyed crowdworkers consider freelancing/self-employment to be their main status (ranging from 66% among macro-workers to 24% for microworkers). One in five combine online platform work with full-time employment (so-called hybrid self-employment). Micro-workers are more likely to consider full-time employment as their main status, while there is also a heavy representation of students in this group.

Online freelancers tend to earn a larger proportion of their income through online platform activities, while 87% of the surveyed microworkers only earn up to 20% of their total income by engaging in crowdwork. Freelancers also tend to work more days in the online platform economy compared to microworkers. It is therefore evident that while online platform work constitutes a secondary activity for most crowdworkers in the sample, those in macrowork/freelancing tend to have more stable or regular engagements, whereas microworkers’ activities are more occasional and circumstantial. This is further reflected in differences in the underlying motivations of the two types of crowdworkers. Most online freelancers do such activities because they want to be their own boss and have greater autonomy over their schedule; in case of microworkers, the main motivation drivers are doing a fruitful activity, earning secondary income, having fun, and killing time. About 1 in 5 crowdworkers are also motivated by the desire to acquire extra technical skills.

One of the main findings of the Crowdlearn project is that there is a “missing training market” in the online platform economy. Platform representatives, but also platform clients, do not see themselves as responsible for workers’ skill development and consider instead that it is the main responsibility of crowdworkers. To ensure that they do not compromise the fuzzy boundary between employees and self-employed persons, most platforms see themselves only as facilitators providing information to workers about skills e.g. what are the most successful workers in the platform doing in terms of projects/tasks; or what skills are in demand by platform clients. They also recommend training courses to their crowdworkers or try to facilitate forums where platform workers can interact and learn from each other.

A notable exception to the above pattern is a platform trying to set up an in-house training capacity. Its main business case is that most of its crowdworkers prefer to learn directly from successful or ‘like-minded’ platform workers, as opposed to what they consider to be distant and irrelevant formal education and training providers (universities, adult learning centers, etc.). Crowdlearn evidence highlights that crowdworkers do not think that traditional learning is valuable as they typically require very fast, advanced and targeted training and learning. Skills needed to do online platform work tend to be very specific and idiosyncratic so crowdworkers heavily rely on modularized, quick and ‘just-in-time’ learning (e.g. internet videos) to acquire them.

A major novelty of the CrowdLearn project is that it measured what types of skills crowdworkers feel are required and have mostly developed through their online platform work. Building on insights from the in-depth crowdworker and stakeholder interviews, the project created a new skills taxonomy, customised to online platform work. Information on hundreds of minor ’in-demand’ skills was first collected (e.g. programming, SEO, English, graphic design, pitching, applying for work, pricing own work, self-presentation, taxes, business permits, learning skills, platform boundary management, time management, work efficiency, handling customers, confidence, independence, discipline). This long array of skills then got translated into a broad categorisation of 10 larger skill groups, namely computer literacy, analytical skills, organisational skills, personal dispositions, communication skills, “being an online worker”, learning skills, “obtaining work on platforms”, foreign languages, and core technical skills. This broader taxonomy was subsequently used to gauge the importance of different skills in the quantitative survey phase.

A key finding of the project is that entering platform work requires workers to already have a very good basic level of technical skills (e.g., graphic design), computer literacy and analytical capability; but that through their work on online platforms, crowdworkers mainly develop their soft skills (personal dispositions, organisational, and communication skills). Unique skills that stand out are those of being an online worker or obtaining work on platforms. These refer to the ability of individuals to navigate efficiently in an online/remote environment, to understand where to find work opportunities and which to spend their time on, how to present and make their profile appealing to potential clients and other such self-branding traits. Significant differences exist however in the skills developed by online freelancers and microworkers; the latter are significantly less likely to develop their communication skills (due to the highly solitary nature of their platform tasks) while they also develop more of their basic digital and analytical skills as part of their crowdwork.

A specific scale — the Workplace Learning in Crowdwork questionnaire — was developed to measure the extent to which platform workers engage in recent workplace learning activities. The Crowdlearn survey reveals that the main workplace learning activities carried out by online platform workers tend to be informal and self-directed. Crowdworkers mostly learn by themselves (by trial and error) or by thinking deeply about their work. The receipt of (client) feedback is also a crucial learning channel.

The analysis has further showed that the nature of platform work tasks is directly related to the intensity of workplace learning. For over 6 in 10 microworkers the tasks they carry out are routine, standardised and decontextualised (i.e. microworkers don’t fully understand why they do them / for whom / what purpose they serve), with little scope for creativity, and this partially explains their lower workplace learning intensity. Online freelancers, by contrast, are more likely to regularly engage in crowdowrk (i.e. a larger number of projects for a greater number of hours) and are more likely to benefit from peer-learning, factors that augment their intensity of workplace learning. For all platform workers, workplace learning activities are highly and positively related to their self-regulatory strategies, namely the extent to which they plan, implement and reflect ex post on their learning.

When the sample is segregated by gender, the analysis reveals that females are less likely to benefit from workplace learning activities in the platform economy. On the other hand, positive workplace learning effects are found for non-native crowdworkers, which reinforces the argument that some marginalised groups on the labour market (e.g., immigrants) could potentially benefit by developing new skills in online platform work.

Another key objective of the Crowdlearn research was to understand how the skills of platform workers are matched in the online platform economy. A main argument typically made in favour of online platforms is that they facilitate skills matching by eliminating coordination costs that exist in traditional labour markets. As intermediaries, platforms typically employ AI algorithms for skills matching, but of concern is their lack of transparency and the use of algorithmic management techniques. The six R’s of algorithmic management — restricting and recommending, recording and rating and replacing and rewarding — are all evident in how skills matching takes place in the online platform economy. Platforms employ algorithmic management to regulate/vet whom they accept on the platform, who is allocated to particular tasks and who gets promoted as a star performer, inter alia. There is thus unequal access and treatment of workers — especially inexperienced newcomers — in some online platforms.

Even after workers overcome entry barriers, significant skills mismatches perpetuate. In online platform markets the most important signaling mechanism of crowdworkers’ skills are the ratings and reputation scores received from clients, as opposed to their formal qualifications. These ratings can be relatively unstructured, lacking context and do not convey constructive feedback for workers to understand how they can improve themselves. Crowdworkers also do not fully understand how reputation scores (the ‘star system’) are used in an algorithmic manner to determine their future probability of success and progression in the platform. Another problem is that many workers feel trapped because they do not have the ability to transfer their skills, knowledge, and experience, to other platforms. The infrastructure of platforms, in terms of proprietary skills ontologies and reputation systems deployed, is such that skills portability is significantly inhibited (‘lock-in effects’).

Based on these findings of the CrowdLearn project, Cedefop has made the following policy recommendations:

• To overcome entry barriers of newcomers, policy makers should collaborate with platforms to experiment with a system of subsidized “micro-internships” in platform work — with such a system, platform clients could be subsidized to hire inexperienced crowdworkers in exchange for the provision of constructive formative and developmental feedback;
• Vocational education and training providers should aim to offer short online courses and workshops that can meet platform workers’ “just-in-time” learning needs and possibly offer high-quality microcredentials;
• Initial education and training systems should focus more on developing people’s self-regulatory learning skills, capabilities and mindsets; while continued focus on digital skills development is needed;
• Policy makers should consider adopting portable portfolio functions to allow workers to display/advertise/transfer qualifications/experiences; they should also engage with platform companies to examine ways of overcoming obstacles to cross-platform portability.