DBA and Startups Act: where is the boundary between self-employed and employment?

What the DBA Act is trying to solve

The DBA Act was introduced on 1 May 2016 as a successor to the VAR. The idea behind this was that the old system left too much room for situations where people were self-employed on paper, while reality rather pointed to employment. This put bogus self-employment at the center, and that is exactly where the law affects many young companies that want to scale up quickly with external capacity.

For startups, this is not a theoretical discussion. In the initial stages, we often work with developers, growth specialists, product consultants or fractional operators who are not on the payroll. This is fine, but only if the employment relationship actually remains out of employment. So the key question is not whether someone sends an invoice, but whether the cooperation materially fits self-employment or is more like working as an employee.

The real key: not the contract label, but the content of the relationship

One of the most important lessons from the development of the DBA Act is that the name of the contract is not decisive. In practice, an assignment agreement can still be regarded as an employment contract if the rights, obligations and actual execution indicate this. In other words, a startup that calls everything “self-employed” is therefore not out of the risk band.

The assessment focuses on the civil employment contract. This looks at whether there is personal work, salary and working “in the service of”, i.e. a relationship of authority. These three elements still form the basis of the assessment. Only when the cooperation at these points remains outside the profile of an employment contract is there room to see the relationship as a real independent effort.

This is important for startups, because young companies are often operationally close to their external parties. The smaller the organization, the faster a freelancer becomes part of the team rhythm, participates in fixed processes and is managed in terms of content. It is precisely this combination that can change the qualification, even if the parties themselves prefer to assume an assignment relationship.

Why authority and embedding are so often the tipping point

In practice, the authority element in particular provides discussion. This is logical, because instructions can also be given with a normal assignment. Not every instruction therefore immediately means employer authority. Rather, the question is what type of management takes place and whether it fits a client-contractor relationship, or an employer-employee relationship.

The case law has further filled in that assessment in recent years. In line with the Deliveroo judgment, all circumstances of the case must be considered in conjunction. Relevant factors include the nature and duration of the work, the way in which working hours and activities are determined, the embedding of the work in the organization, the obligation to perform the work, the method of remuneration, the commercial risk and whether someone acts like an entrepreneur in business. So there is no trick that is decisive in itself.

For startups, this means that “embedding” in particular requires attention. If someone structurally participates in the core of your product, at fixed times, led by internal leads, in exactly the same operation as employees, then it becomes more difficult to position that collaboration as a pure independent effort. That risk further increases if someone has little commercial risk and hardly any space to really design the assignment independently.

Entrepreneurship helps, but is not a panacea

At the same time, the image isn't black and white. The assessment can also take into account whether someone actually acts as an entrepreneur outside the specific assignment. Think of reputation building, acquisition, the number of clients and the duration of the connection with a specific client. That entrepreneurship can therefore play a role in qualification.

But this is also where an implementation problem arises for clients. The document shows that it is more difficult for clients than for the tax authorities to properly determine which aspects of external entrepreneurship are relevant and what weight they have. This is an important insight for startups: it is nice if a freelancer also works for others, has his own profile and makes an acquisition himself, but that does not automatically remove the risk if the actual collaboration is otherwise very similar to employment.

What model agreements do and don't do

Around the DBA Act, a lot of attention has long been paid to model agreements. The system behind this is that the client and contractor jointly draw up an agreement that should reflect the actual working method. Such an agreement can be submitted to the tax authorities and can provide certainty about payroll taxes. But that certainty only applies within the limits of what has actually been agreed and implemented.

That is immediately the limitation for startups. A model agreement is not a shield if daily practice differs from it. If someone ultimately works as part of your team, under your direction and within your permanent organization, it can still be concluded afterwards that there is an employment relationship. In that case, the tax authorities can impose a correction obligation or additional tax assessment on the client.

In addition, the practice of model agreements itself has changed. As of 1 January 2024, the model agreements based on free replacement were terminated. And since September 6, 2024, the tax authorities no longer process new applications or extensions of model agreements. Current, approved model agreements will be respected until December 31, 2029, as long as they remain in line with legislation and case law, but that is therefore not carte blanche either.

For a startup, the practical conclusion is therefore quite sober: a good contract helps, but only if the contract follows reality. Not the other way around.

Enforcement since 1 January 2025: why this is now extra relevant

For years, the DBA Act was notorious for unclear enforcement and a long enforcement moratorium. That has changed. As of 1 January 2025, the focus of enforcement on employment relationships will focus on clients' payroll taxes. So the risk lies more emphatically with the party that hires, and for this blog, these are precisely the startups and scale-ups that want to scale up flexibly.

According to the piece, there will be a soft landing before 2025: no administrative fines will be imposed in that year, even in the event of malice or obvious bogus self-employment. That sounds mild, but it doesn't mean nothing can happen. Correction obligations and additional payroll taxes can indeed be imposed. In addition, the correction will in principle work back to 1 January 2025, except in the event of malice or if an earlier instruction has not been followed.

This topic remains relevant for 2026 as well. The document states that the normal penalty rules would in principle apply again from 1 January 2026, but that after debate in December 2025, the soft landing was partially extended: no absenteeism fines in 2026, but offence fines for employers and employees. In particular, this shows that the direction of enforcement is no longer optional.

The financial consequences of requalification can be substantial

If a self-employed person is nevertheless seen as an employee afterwards, it is not just theoretical relabelling. In concrete terms, the document states that rights to pension and employee insurance can then arise, provided that no contributions had been paid before. In such a case, a pension fund can still collect the pension contribution from the employer. An additional tax can also follow for employee insurance, with a retroactive effect of up to five years.

For startups, this is exactly why a DBA analysis is not just a tax or legal formality. In a growth company, several external parties can be hired in a similar way at the same time. If one collaboration is already close to employment, it can therefore have a wider effect on your organization, processes and cost structure. This is not just about the individual contractor, but about how your entire flexible shell is designed.

The web module is a tool, not a safe haven

Because practice needs more clarity, the web module was also developed as a tool for qualifying employment relationships. But the piece is clear about that: the outcome of the web module remains an indication, and the actual situation remains decisive. In addition, the web module cannot be used when hiring a legal entity. This is a relevant limitation, especially for startups that regularly contract through a personal holding or management company.

In practical terms, this means that you don't have to drive too much comfort from tools, templates or standard documents. They can help with structuring, but ultimately, the concrete structure of the collaboration counts. Operational practice prevails about paper reality.

What startups should take away from this today

For founders, the first question is not: can we hire someone as a self-employed person? The better question is: how is this person actually going to work within our company? As soon as someone is structurally embedded, bears little independent entrepreneurial risk, has to perform personally and is strongly managed in terms of content, the DBA discussion becomes serious.

In addition, it is wise to design cooperation not only legally but also operationally. A contract that exudes independence, while daily practice points to employment, creates false security. And that is exactly what the DBA Act focuses on.

Finally, it is good to note that the legislator is still looking for further clarification. The document describes that a bill clarifying the assessment of employment relationships and legal presumption is now in the pipeline, but also that the effect is still evolving. For startups, this is primarily a signal not to wait for “the final new rules”, but to already look carefully at existing collaborations.

Conclusion: flexibility is fine, non-commitment is not

The DBA Act does not prohibit startups from working with self-employed workers. However, the law does force you to look honestly at the content of the cooperation. Not the invoice, not the job title and not the standard template are decisive, but whether someone really works at their own expense and risk, or is essentially part of your organization.

For startups, that may be a little less sexy than “build fast”, but it is legally much healthier. If you focus on authority, embedding, business risk and actual implementation at an early stage, flexibility prevents flexibility from later turning into additional taxes, insurance questions and pension claims. And that is exactly why the DBA Act now belongs on the agenda of founders, cFos, legal and people operations.

‍