In this article
A flex-staffing app feels like the opposite of an employment contract: you book an hour, pay a rate, and stop thinking about it. The EU Platform Work Directive flips that from 2 December 2026 — seven numbers show how close that freelancer already sits to your employee, and what it costs you if a court or inspector agrees.
Every independent restaurant knows the pattern by now: a Saturday fills up, someone calls in sick, or the terrace suddenly needs two extra pairs of hands — and instead of phoning around, you open an app. Temper, YoungOnes and their national equivalents let you book a freelance bartender, waiter or kitchen porter for exactly one shift within minutes. You pay an hourly rate, the app handles the rest, and there is no contract, no payroll entry and no notice period to think about. For an independent venue with no HR department, this has quietly become the default valve for any peak the core team can't absorb.
That exact convenience is what the EU is now targeting. The Platform Work Directive (EU) 2024/2831 entered into force on 1 December 2024, and every member state must have it transposed into national law by 2 December 2026 — just over three months from today. At its core sits a rebuttable presumption of employment: once a platform relationship shows signs of "control and direction" — the platform sets the rate, fixes shift times, tracks performance through an app, restricts working for competitors — the burden of proof flips. It is no longer the freelancer who must prove they are really an employee; the platform (or, in practice, the venue directing the shift) must prove they genuinely are not.
This is not a theoretical dossier. The Dutch Supreme Court delivered the precedent this directive now makes repeatable EU-wide back in 2023: Deliveroo riders, treated as self-employed "partners" for years, were recognised as employees, with mandatory collective-agreement terms and pension contributions applied retroactively. The European Commission's own impact assessment estimates that 5.5 million of the 28 million people working through platforms across the EU are currently misclassified — 93% of them filed as self-employed. A restaurant that books the same Temper freelancer for the same shift, week after week, directed exactly like a permanent staff member, sits closer to that statistic than most owners realise.
This article walks through the seven numbers that make the dossier concrete: the deadline, the original test, the scale of the platform landscape, and the precedent that already exists. It's followed by a calculator that translates the risk into your own flex-staffing spend, and a concrete action plan for what you can do this week — without having to drop flex-staffing apps altogether.
Why this reaches every restaurant, not just delivery platforms
The Deliveroo rulings were about bicycle couriers, and it's tempting to assume this dossier stays confined to delivery apps. That's a misreading the directive itself explicitly corrects: it applies to any "digital labour platform" that organises work and exercises control, regardless of sector. An app that matches freelance hospitality staff fits that definition the moment it does more than connect supply and demand — the moment it fixes rates, rates performance, or sets the terms of the assignment.
Your restaurant isn't a platform itself, and that doesn't automatically protect you. The directive targets the employment relationship itself, and in practice that is the venue directing the freelancer on the floor: who sets the start time, which tables they cover, what uniform they wear, and whether they get booked again next time. An employment tribunal or labour inspector looks at the facts of that relationship, not at the fact that payment happened to run through an app. Which is exactly why the question shifts from "is the app the employer?" to "does the venue behave like one?" — a question that applies on any floor where a flex-staffing app is used as a structural fix rather than an occasional one.
"I just use an app, I don't have a contract" is precisely the assumption the directive undermines. A contract was never the test — it was always the actual relationship, and that's exactly why so many owners are unknowingly exposed here. They assume the absence of paperwork protects them, when it's the paperless, app-directed relationship itself that triggers the presumption of employment.
Guide Everything about restaurant staff From hiring to departure — the complete guide Read the guideSeven numbers that make the dossier concrete
Each number below stands on its own, but together they map the timeline, the test, and the scale of a dossier that is already running.
1. 2 December 2026 — the date the directive becomes national law
The Platform Work Directive isn't a regulation that applies directly, the way the EU packaging law (PPWR) does — it's a directive, which means every member state has to transpose it into its own legislation first. That transposition must be complete by 2 December 2026, exactly two years after the directive entered into force on 1 December 2024. Some member states already have draft legislation on the table; others haven't started — but the deadline itself is fixed and not up for negotiation.
That doesn't mean nothing changes before then. National courts already apply existing employment tests today — as the Dutch Supreme Court did in 2023, well before the directive existed — and the directive largely confirms and reinforces that direction rather than starting from zero. By the time the law formally changes, case law in many countries will already have moved a long way in the same direction.
Four moments frame the whole directive: the first proposal, adoption, entry into force, and the transposition deadline. Today, late August 2026, you sit well along the way between the last two.
The "today" marker's position is calculated from this article's publication date and won't move on its own if you're reading this later — count forward from 1 December 2024 yourself if you want the exact time remaining to 2 December 2026.
2. 5 — the original control indicators, still the compass
The Commission's first proposal in 2021 worked with a concrete test: the presumption of employment applied once at least two of five indicators were present. Those five were: the platform sets or caps pay; the platform electronically monitors or assesses performance; the platform imposes binding rules on appearance, client contact, or how the work must be carried out; the platform restricts freedom to choose working hours or to switch the app off; and the platform forbids or discourages working for others.
That fixed "two of five" threshold was dropped from the final text — each member state may now decide for itself which facts trigger the presumption, while taking national law, collective agreements and existing EU case law into account. That reads like a weakening, but in practice it mostly means the original five indicators remain the reference point national legislators and courts are drafting and ruling from. Anyone who checks their own use of a flex-staffing app against those five will see immediately where the exposure sits.
Tick what applies to your venue. This isn't a legal test, but it shows how close you sit to the original "two of five" threshold.
Tick the signals that apply to your venue.
Based on the five indicators from the Commission's 2021 proposal — still the reference point for most national transposition bills, even though the exact threshold varies by country.
3. 500+ — the number of digital labour platforms active in the EU
Platform work is no longer a niche phenomenon. The European Commission's impact assessment counted more than 500 digital labour platforms active across the Union, and hospitality and logistics together form a large share of that landscape. Temper and YoungOnes are the best-known names for hospitality staffing in the Benelux, but they're two of hundreds of comparable operators trading under other brands in other member states.
What those 500+ platforms have in common is that they all fall under the same definition of "digital labour platform" the moment they organise work through algorithms or standardised processes. The directive makes no exception for a platform that "only" arranges four-hour shifts rather than delivery runs — the legal test is the same, regardless of how small the individual assignment is.
Four figures from the European Commission's own impact assessment set the context: this isn't a niche dossier about delivery riders, it's a structural part of how Europe staffs itself today.
Source: European Commission impact assessment accompanying the directive proposal (SWD(2021)397) and the final text of Directive (EU) 2024/2831.
4. 28 million — the number of people working through a platform across the EU
The Commission estimates that 28 million people across the EU work through a digital labour platform at some point, and that number is growing faster than the labour market as a whole. That makes platform work a structural part of how Europe staffs itself, not a temporary side income for students. For hospitality specifically, it means the flex-app freelancer is becoming an increasingly fixed link in how a venue plans staffing around peaks — exactly the kind of structural, recurring use that trips a control test fastest.
The more structurally a venue leans on a platform to fill its baseline staffing, the more likely an individual booking stops looking like occasional help and starts looking like a disguised permanent shift — and that distinction is exactly what a court or inspector examines.
5. 5.5 million — the Commission's own estimate of how many are misclassified
Of those 28 million platform workers, the Commission estimates 5.5 million are currently misclassified as self-employed — people who, based on the actual facts of their working relationship, should be employees. That figure isn't a forecast of how many will eventually go wrong; it's the Commission's own estimate of how many are wrong right now, waiting for the first inspection or dispute to bring the question in front of a court or labour authority.
This is precisely the population the directive targets, and there is no reason to assume hospitality is under-represented in it. A venue that has never checked its own Temper rota against the five indicators genuinely doesn't know whether its own freelancers sit inside that 5.5 million — until the moment someone actually checks.
6. 93% — the share of platform workers currently filed as self-employed
93% of all platform workers across the EU are currently booked as self-employed. That figure is why the presumption of employment is such a heavy intervention: it flips the default assumption of an entire sector. Previously, a worker had to prove they were actually employed; now the platform (or the venue directing the assignment) has to prove the freelancer genuinely isn't.
For a restaurant owner, this number mostly means "most people on these apps are genuinely self-employed anyway" is no longer a reassurance — it's exactly the assumption the directive was written to re-examine. The norm shifts from self-employed-unless-proven-otherwise to employee-unless-successfully-rebutted.
7. 2023 — the year a Dutch court already showed what this costs
Before any EU directive existed, the Dutch Supreme Court ruled on 24 March 2023 in a case that shows exactly what a successful presumption of employment costs in practice. Deliveroo riders, treated by the company as self-employed "partners" for years, were recognised as employees based on the classic tests of work, pay and authority — not on how the contract described itself, but on how the relationship actually functioned.
A follow-up ruling in November 2023 confirmed the consequences: Deliveroo had to apply the applicable collective agreement for professional goods transport and pay pension contributions retroactively for the period during which the riders were already considered employees. That is the precedent the EU directive now makes repeatable across the whole Union — not as an exception for delivery platforms, but as the model every national transposition is drafting against when deciding exactly when the presumption of employment kicks in.
What would reclassification cost you?
The figures above are European. This calculator translates the risk into your own flex-staffing spend — not legal advice, but a planning number that makes the abstract dossier concrete.
The model is deliberately simple: how much more would you pay per month if your current flex-staffing spend suddenly fell under a normal payroll burden, and what would that cost retroactively over the period a dispute or inspection could reach back into? Both rates are editable — neither is a fixed legal fact, since the actual payroll overhead and the look-back period both vary sharply by country and by circumstance.
Run your own risk
Enter your monthly flex-staffing spend, along with a payroll-overhead percentage and a look-back period you estimate yourself.
—
Defaults: €3,000 a month, 25% extra payroll overhead, a 2-year look-back — adjust them to your own situation.
Not legal or tax advice. The actual payroll overhead and limitation period depend on your country, your contract type, and the precise national transposition of the directive.
Notice that the third tile — months of your own spend — is independent of how much you actually spend each month. It depends only on the overhead percentage and the look-back window: double the look-back period and that ratio doubles too, whether you spend €500 or €5,000 a month on flex-staffing apps.
This is a planning number, not legal or tax advice. The actual payroll overhead, the limitation period for reclassification, and exactly how your national transposition defines the presumption of employment all vary by country and by case. Discuss your own situation with your accountant or a labour lawyer before drawing conclusions about your own rota practice.
What you can do this week
You don't have to stop using flex-staffing apps to reduce this risk — you mainly need to know where your own use scores against the five signals, and have an alternative ready for the cases closest to a permanent shift.
Check your own use against the 5 signals
- Run your most recurring Temper or YoungOnes bookings through the checklist above — the same person, the same shift, every week, is the pattern that triggers scrutiny fastest.
- Where possible, let the freelancer propose their own rate instead of setting a fixed internal one.
- Avoid a mandatory rating that counts toward future bookings — ask for informal feedback instead if you need it.
- Revisit uniform and script requirements: a visible difference in presentation between freelancer and permanent staff protects the self-employed status, not the other way round.
Build your own rota as the alternative for structural gaps
- Use the free rota and staff-planning tool to plan recurring peaks ahead of time instead of booking the same freelancer every week.
- Invest in cross-training so permanent staff can absorb a peak without outside help.
- Plan seasonal staff structurally ahead with the staffing curve instead of reactively topping up.
- Compare the real cost of a temp-agency worker — there, the employment relationship is already clear, because the agency is the employer.
Track your own country's transposition
- Check with your accountant or hospitality federation when the transposition law is expected in your country — timing varies sharply between member states.
- Ask specifically which facts your national law will treat as a control indicator, rather than relying on the original "two of five" test.
- Document, for your longest-running freelancers, why the relationship remains genuinely self-employed — that documentation is exactly what shifts the burden of proof if it's ever tested.
The bottom line
A flex-staffing app remains a legitimate and useful tool for absorbing peaks — the directive doesn't ban anything, it only shifts the burden of proof once a relationship starts looking too much like a permanent shift. The difference between a freelancer who occasionally helps out and a disguised employee isn't the app that handles payment, but who sets the rate, who fixes the shift, and who decides whether there's a next time.
By 2 December 2026, that distinction becomes legally enforceable everywhere in the EU, with the burden of proof on you rather than on the freelancer. The Dutch Deliveroo case already showed what that costs in practice: collective-agreement terms and pension contributions, applied retroactively to a relationship that had been treated as self-employed for years.
The seven numbers above aren't a reason to drop flex-staffing apps — they're a reason to check your most recurring bookings against the five signals now, before an inspection or a dispute makes that choice for you.