Strengthening Individualised Support for Jobseekers Furthest from the Labour Market in Greece

1. Assessment and recommendations

Abstract

Greece’s Public Employment Service DYPA plays a crucial role in connecting vulnerable groups to jobs in Greece, despite limited financial resources and staff. This chapter presents the main assessments and recommendations on how digital tools and complementary services can help DYPA enhance support for vulnerable jobseekers. In particular, the current profiling process could be strengthened by making better use of available data and improving data coverage, so that vulnerable jobseekers are more reliably identified. In addition to flagging jobseekers who need intensive support, a revised tool could help counsellors to better understand individual barriers to employment, develop tailored individual action plans, and propose referrals to services that can, where appropriate, be contracted out. Effective implementation of these functionalities will require close alignment with DYPA’s ongoing review of its profiling approach, strict compliance with data protection and fairness standards, and robust monitoring and evaluation.

1.1. High caseloads limit DYPA counsellors’ ability to provide individualised support to vulnerable jobseekers

Jobseekers further from the labour market often face multiple and complex barriers to employment (for example, related to skills, experience, health or care responsibilities). These jobseekers (also referred to as vulnerable jobseekers in this report) require therefore intensive, individualised and co‑ordinated support for a successful labour market integration. Providing such support requires a sufficient number of job counsellors. DYPA has made a lot of progress in decreasing the caseloads over the past years by employing more counsellors and successfully helping more jobseekers into employment. Despite the good progress made, the caseloads remain still significantly higher than in many other OECD countries. DYPA employment counsellors currently manage caseloads of around 600 jobseekers on average, while the caseloads can be even higher seasonally and regionally. Against this background, DYPA should continue to invest in the digitalisation and modernisation of its processes. Well-designed digital tools can streamline administrative processes and enhance the efficiency and effectiveness of counsellors’ work, thereby freeing up time and resources to focus on those furthest from the labour market and deliver more individualised support. To maximise the impact of digitalisation, it is important that the development of new digital tools is strategically co‑ordinated both within DYPA and across relevant institutions. In particular for vulnerable jobseekers (officially referred to as “special and vulnerable groups” by DYPA), the different types of support provided through these tools should be accessible via a single entry point, making services easier to navigate and reducing the complexity of the interactions with DYPA.

1.2. The current profiling tool should be made more data-driven to meaningfully identify jobseekers in need of additional support

The current profiling tool relies primarily on information collected through a questionnaire and assigns jobseekers to five categories based on their job readiness. In practice, a large share of jobseekers is assigned to the middle category (medium distance from the labour market), limiting the tool’s ability to meaningfully differentiate jobseekers according to the level of support they require. There is significant scope to reinforce the current profiling process by making it more data-driven, that is by leveraging historical jobseeker data and applying modern statistical techniques to identify clients requiring individualised and holistic support.

DYPA could complement the information gathered through the profiling questionnaire with administrative data on unemployment and employment histories, benefit receipt and other relevant characteristics. The analysis of DYPA’s rich jobseeker microdata shows that several demographic and labour market characteristics – such as age, gender, level of education, the number of previous unemployment periods, and time since the end of the last unemployment period – are highly predictive of distance from the labour market. Similarly, data on previous employment records and job type from the employment register (ERGANI) contain strong predictors for long-term unemployment. Much of this information is already available in DYPA’s Integrated Information System (IIS) and could be readily leveraged to better identify vulnerable jobseekers. Coverage could be further enhanced by extending the existing data exchange with ERGANI to systematically include additional characteristics of previous employment that are not yet available in DYPA’s IIS.

Incorporating this rich information into profiling would increase the ability of the tool to segment jobseekers accurately according to their distance from the labour market and to meaningfully distinguish between clients of low, medium and high risk, avoiding over-concentration in the middle category. A more data-driven profiling tool would also enable DYPA to focus and prioritise counsellor meetings on those furthest from the labour market and most likely to benefit from additional support, while delaying meetings with jobseekers who are highly employable.

1.3. Despite being compulsory, the profiling process is not always completed for the more vulnerable groups

Although the profiling process is compulsory for most jobseekers, completion rates remain sub-optimal, especially among certain vulnerable groups, undermining the usefulness of profiling precisely for those most in need of support. Some vulnerable jobseekers are more likely than others to end up without a score because key variables in the profiling questionnaire are missing, making it impossible to calculate the score. The analysis of the data shared by DYPA shows that while on average around one in two jobseekers completes the profiling questionnaire and receive a suggested profiling category from the tool, this share is lower for some categories of vulnerable jobseekers, such as young offenders, refugees and other immigrants. Some vulnerable jobseekers face barriers to completing the profiling questionnaire online altogether, such as difficulties related to language, literacy and digital skills. Incomplete or inaccurate information increases the risk of misclassification and weakens DYPA’s ability to systematically identify and prioritise those furthest from the labour market.

To address this, DYPA should aim to increase the share of jobseekers who complete the profiling questionnaire and receive a final profiling category, with particular attention to groups with low completion rates. To reduce data gaps and missing values in the variables required to calculate the score, DYPA could streamline the questionnaire to focus on the most relevant information, while linking this, where appropriate, with data already available in DYPA’s IIS or other administrative registries. At the same time, the profiling process should be made more user-friendly, with clear guidance and targeted support for those unable to complete it independently.

Moreover, DYPA could flag potentially vulnerable jobseekers even when a profiling score cannot be generated. In these cases, the segmentation into a risk category could be carried out using the data already available in DYPA systems at registration, such as prior long unemployment periods, disability status, release from prison, and for people for whom the data from ERGANI are available, also employment history. This complementary approach would reduce the risk that vulnerable jobseekers are overlooked and help counsellors provide timely support.

1.4. A revised digital profiling tool can help counsellors better understand jobseekers’ barriers and design tailored action plans

Beyond simply flagging vulnerable jobseekers who need intensive support, a revised profiling tool could provide richer insight into the specific barriers each individual faces. Rather than only producing a profiling score, the tool could make explicit which factors are most strongly associated with a lower probability of finding work. The analysis of the linked data from DYPA and ERGANI shows that these factors include, among others, occupation, the reason for termination of previous employment and educational attainment. This would shift profiling from a pure classification exercise towards a more transparent, explainable and informative decision-support tool.

In practice, this could take the form of a digital interactive dashboard that highlights jobseekers at high risk of long-term unemployment and displays the key drivers of that risk. To be effective, such a tool should be fully integrated into counsellors’ existing interfaces and workflows, so that risk indicators and suggested actions are visible and easy to use for counsellors. Building on the information provided by the tool, counsellors can develop well-tailored individual action plans, combining appropriate measures and referrals in ways that can more effectively support vulnerable jobseekers into sustainable employment.

1.5. Digital tools can support the co‑ordination of services for vulnerable clients

Vulnerable jobseekers often need support that extends well beyond employment services, involving support that cuts across different institutions such as other ministries, agencies, municipalities and NGOs. While DYPA counsellors sometimes inform clients about services available in their area that are not directly provided by DYPA, this currently happens in an informal and ad hoc way. It relies heavily on individual counsellors’ knowledge, which can vary considerably between offices and regions. As a result, access to complementary support – such as housing, health, childcare or legal advice – may be uneven and dependent on chance rather than need.

Digital tools offer an opportunity to make this process more systematic and equitable. DYPA has already taken important steps by signing Memoranda of Understanding with organisations representing the interests of vulnerable jobseekers. Building on this, a comprehensive mapping of services available to vulnerable clients at regional and local level could be developed and embedded in the digital tool supporting jobseeker profiling. Employment counsellors could then access a directory of available services directly within their interface and use it to recommend appropriate support options to jobseekers as relevant. This would help to streamline and standardise the process of making clients aware of non-DYPA services, strengthen co‑operation with partner organisations and, ultimately, provide more co‑ordinated and holistic support to vulnerable clients.

1.6. New functionalities to better support vulnerable jobseekers should be considered within DYPA’s ongoing review of profiling processes

DYPA is already in the process of evaluating and revising its current profiling tool. Any effort to develop new digital solutions to better identify and support vulnerable clients should therefore be closely aligned with this broader review. Developing separate or parallel tools risks duplication, inconsistency and additional complexity for counsellors. Instead, DYPA should assess the feasibility of incorporating the functionalities proposed in this report to enhance support for vulnerable jobseekers into the revision of the overall profiling tool. Doing so would help ensure that the revised tool serves both general and vulnerable jobseekers in a consistent way.

1.7. Relevant stakeholders should be involved early in the development process of the new digital profiling tool

It is essential to involve all relevant stakeholders from the earliest stages of designing and developing the new digital tool. Early engagement helps ensure that the tool responds to real needs rather than being driven solely by technical possibilities. Once DYPA has identified the key stakeholders – such as counsellors, vulnerable jobseekers, IT staff and partner organisations – it could organise structured discussions to co-develop a theory of change and related results chain. The theory of change and results chain can guide both the design of the tool and the establishment of a monitoring and evaluation framework. Such a participatory process not only improves the relevance and usability of the tool but also strengthens ownership among those who will use it in practice. This, in turn, increases the likelihood that the tool will effectively support vulnerable jobseekers.

1.8. Profiling must comply with data protection, transparency and fairness standards

The proposed tool will require the collection, linkage and analysis of personal data on jobseekers, potentially including sensitive information and data relating to particularly vulnerable individuals. It is therefore essential that the design and use of the tool will fully comply with Greek data protection regulations, the General Data Protection Regulation (GDPR), and broader standards of transparency, fairness and accountability. This becomes even more important if artificial intelligence is used for some of the tool’s functionalities, as this may raise additional concerns around bias, explainability and automated decision making.

In practice, DYPA should ensure that jobseekers are clearly informed, at the point of data collection, about what information is being gathered, how it will be used, and what safeguards are in place. Where appropriate, individuals should be offered the possibility to opt out of certain uses of their data without jeopardising their access to core services. Documentation of the tool – including its underlying logic, key variables and safeguards – should be accessible to both counsellors and clients. Taking a proactive and systematic approach to data protection and transparency will help build trust in the new profiling tool and is particularly important for vulnerable groups, who may be more exposed to risks arising from the misuse of their data.

1.9. Consolidate partnerships and strengthen data exchanges with other institutions

The quality and completeness of the information feeding into the profiling tool will directly determine the accuracy of its outputs. If relevant jobseekers’ information is missing, the tool will struggle to accurately assess their distance from the labour market and the type of support they require. It is therefore crucial to establish protocols that allow DYPA to retrieve the necessary data from the institutions and registries that hold these, within a clear legal and data protection framework. For example, the analysis of DYPA data reveals that jobseekers identified as belonging to a special social group – a sub-category of vulnerable jobseekers – are around five times more likely to have missing information on their education.

The web services made available by the Ministry of Digital Governance provide a solid basis for strengthening data exchange between DYPA and other institutions. DYPA already makes use of these web services, for example to retrieve information from the employment register ERGANI on salaried employment. Data exchange could be expanded to other institutions too, such as the social security (EFKA) which also include information on self-employment, so to obtain a more comprehensive picture of individuals’ labour market histories.

1.10. While the proposed tool can help streamline processes, it is essential that human judgement remains central

The proposed digital tool is intended to complement and enhance the work of employment counsellors, not replace them, and its recommendations should be understood as support to counsellors’ decisions. This is particularly important for vulnerable jobseekers, whose situations are often complex and may not be fully captured in available data. Counsellors need to critically assess any suggestions generated by the tool, taking into account the wider context of the jobseeker’s circumstances, and validate or adjust them as appropriate. Maintaining this “human-in-the-loop” approach helps ensure that support remains person-centred and responsive.

If, in some cases, decisions were to be taken on the basis of automated outputs without individual counsellor validation, this would need to be handled with great care. Jobseekers should be clearly informed when automated processing plays a significant role in decisions affecting them and, where required, their explicit consent should be obtained. Transparency about how the tool works, the role it plays in the support process and the continued importance of human judgement will be critical to building and maintaining trust in the tool.

1.11. Face-to-face interactions remain essential

For some vulnerable clients, providing the information required by the tool through digital interfaces may be challenging, whether due to limited digital literacy, lack of access to IT equipment, or broader barriers such as language difficulties or low confidence in using online tools. Relying exclusively on digital channels risks leaving behind precisely those who are most in need of support. To ensure inclusiveness and effective outreach, it is therefore crucial that vulnerable jobseekers also have the opportunity to seek support in person.

In Greece, this is already the case in some areas through EKO offices, which are specifically designed to receive and accompany vulnerable clients. However, these offices are currently limited to a small number of locations. To provide effective support across all Greek regions and improve outreach, this type of face‑to-face provision should be scaled up, either by expanding the network of specialised EKO offices or by embedding similar functions within existing KPA2 offices.

1.12. Put a monitoring and evaluation framework in place from the outset

To ensure that the tool delivers on its objective, it is essential to have a well-defined monitoring and evaluation (M&E) framework in place from the very beginning. Without such a framework, it would be difficult to know whether the tool is being implemented as intended, whether it is being used consistently by counsellors, and whether it is actually improving outcomes for vulnerable clients.

A robust M&E framework enables continuous tracking of implementation, supports the early detection of deviations from the original design, and provides a sound basis for assessing whether the tool’s impact aligns with its strategic objectives. In doing so, it plays a key role in governance: it allows policymakers and project managers to systematically monitor progress, ensure accountability for results, and promote transparency vis-à-vis internal and external stakeholders.

1.13. Capacity constraints can be alleviated by contracting out some services to external providers

One flexible way to address the capacity constraints of DYPA is to contract out some services to external providers. Carefully designed contracted-out provision can expand the range and intensity of support available to vulnerable jobseekers without overburdening DYPA’s structures. Given the diversity of needs of DYPA’s vulnerable jobseekers, programmes could be separately designed to cater to individual needs.

DYPA could design a contracted-out programme for long-term unemployed individuals, offering continued engagement with private employment counsellors to address barriers to employment. This would offer additional payments for sustained employment by participants, to incentivise more durable job matches.

Separately, an additional programme could be designed to help people with health constraints or disabilities. Greater weight could be given to tenders from external providers who can demonstrate how their service would link in with broader local social services. This would help to ensure that for people with complex needs, local services are joined up to holistically address barriers. Continued employment support post-job entry – working directly with jobseekers and employers, and offering additional support for workplace adjustments – would help to sustain job placements, particularly soon after entry into work.

Key policy recommendations

Make profiling more data-driven and inclusive

• Enhance the current profiling tool used to flag vulnerable jobseekers by leveraging historical jobseeker data and applying modern statistical techniques.

• Improve the accuracy of the profiling tool by combining questionnaire responses with administrative data on unemployment and employment histories.

• Formalise and expand data-sharing protocols with key institutions (such as the Ministry of Labour and EFKA), using the web services of the Ministry of Digital Governance, to obtain more complete labour market histories and strengthen profiling accuracy.

• Increase the share of jobseekers who receive a profiling score – with particular attention to vulnerable groups with low completion rates, such as refugees, young offenders and former prisoners – by streamlining the questionnaire, reducing missing values and providing targeted support to complete it.

Use digital tools to tailor support and co‑ordinate services for vulnerable clients

• Use data to identify the main drivers of distance from the labour market and display them in an interactive digital dashboard to support counsellors in designing tailored individual action plans.

• Map available local services (e.g. housing, healthcare, childcare, legal support) provided by other organisations and institutions that are needed to deliver holistic support to vulnerable jobseekers.

• Integrate this service mapping into the counsellor interface, enabling counsellors to recommend and refer clients to relevant services in a structured and consistent way.

Safeguard data protection, transparency and fairness in profiling

• Ensure that the design and use of the tool fully comply with the Greek data protection law and the GDPR, including providing clear information to jobseekers on what data are collected, how they are used and what safeguards are in place.

• Clearly communicate to jobseekers the respective roles of automated processing and counsellor judgement in decisions that affect their support and obligations.

• Where appropriate, allow jobseekers to opt out of certain uses of their data without jeopardising their access to core services.

• Document and make accessible the tool’s underlying logic, key variables and safeguards and, in particular where artificial intelligence is used, regularly monitor for potential bias.

Keep human judgement and inclusive access at the centre of service delivery

• Maintain human oversight in which counsellors remain responsible for final decisions, using the tool as decision-support.

• Ensure that digital tools targeted to vulnerable jobseekers are accessible through a single entry point, to simplify navigation and reduce complexity.

• Scale up face‑to-face provision for vulnerable clients by expanding the network of specialised EKO offices or embedding similar functions within KPA2 offices, ensuring that vulnerable jobseekers can complete the profiling process also in person.

Embed monitoring, evaluation and stakeholder engagement from the outset

• Involve key stakeholders – including DYPA counsellors, vulnerable jobseekers, IT staff and partner organisations – early in the design process to co-develop a theory of change and results chain for the new tool.

• Establish a robust monitoring and evaluation framework from the start, with indicators on tool usage and outcomes for vulnerable jobseekers and use the findings to adjust both the tool and associated services over time.

Design contracted-out services to flexibly add capacity to support vulnerable jobseekers

• Design the parameters for pilots to test out two new contracted-out employment programmes. One programme should cater to the long-term unemployed and another to those with health conditions. Design should include considerations on referral mechanism, programme length, service provision, minimum service standards and performance‑related payment structure.

• Establish a business case for piloting the new contracted-out intervention to provide the rationale and secure adequate funding.

• Appoint a senior responsible officer and convene a policy team to manage implementation, and select participating KPA2 offices. Initial market testing should be used to test viability, and data analysis conducted to establish baselines for setting performance‑related payments.

• Embed evaluation directly into pilot design and delivery. A process evaluation will provide qualitative information on implementation, to improve programme design. Alongside this a counterfactual impact evaluation will permit analysis of the impacts of the programmes for the vulnerable jobseekers.