Service · Software development

Data Act: data access and switching.

The Data Act gives whoever uses a device the right to the data it produces, and requires cloud services to make switching away easy. Both affect your architecture before they affect your legal department. We build the integration, the permissions layer and the export route that make this possible.

If you are mainly looking for the front end, where users request their own data and third parties are granted or lose consent, take a look at our data access portal for connected products.

User accessSharing with third partiesExit arrangementPermissions layer

What the regulation changes, and where the work lies.

The Data Act (Regulation (EU) 2023/2854) has applied since 12 September 2025. It governs two things that at first glance seem unrelated. First: anyone who uses a connected product has the right to the data it produces, and can also have a third party request that data. Second: anyone who purchases a cloud service must be able to leave it without unnecessary barriers. For products placed on the market from 12 September 2026, they must also be designed so that this data is accessible by default.

The practical question underneath is who owns what a device knows. Until now, that was implicitly the manufacturer, because it sent the data to its own platform. The regulation turns that into a right for whoever uses the device, and that affects revenue models that relied on that exclusivity. For suppliers this is not a disaster, but it is a choice: selling maintenance because you are the only one who sees the measurement data is different from selling maintenance because you know best what it means. The first disappears; the second does not.

We do not do the legal work – that is what you have a lawyer for – but we do everything beneath it: an integration through which data can be accessed without anyone needing to get into your system, a permissions layer that determines who may request what and how shared access is revoked again, and an export route that is genuinely usable. This often touches on integrations with other systems, IoT platforms and cloud exit projects.

We do this on its own and as part of building custom software. If it concerns equipment in the field, look at a custom IoT platform; if the question is where your data is stored, look at sovereign cloud and cloud exit. How we handle data ourselves is set out in our information security policy.

Three types of Data Act work we take on.

Which part affects you depends on what you supply: a device, a cloud service, or both. In the first conversation, we determine which of the three needs to come first.

First engagement · fixed sprint budget

Access for the user themselves

A route through which whoever uses the product can reach their own data: measured values, states and events, in a common format and without anyone from your side having to be involved. For this, we separate the raw layer from derived insights, because that boundary determines what you must share and what remains yours. Anyone who only makes that distinction when a request arrives has to pull their system apart.

Raw layer kept separateCommon formatSelf-serviceTime series
Mid-sized project · fixed sprint budget

Sharing with third parties, with permissions and revocation

Users may have a third party request their data, such as a maintenance provider, an insurer or one of your competitors. That calls for a permissions layer: who may access what, for how long, and how the user withdraws that access again. It also needs logging, so that access can be audited rather than merely possible. This is also where the protection of your trade secrets takes shape, because the regulation leaves them intact.

PermissionsRevocationAccess logTrade secrets
Larger project · fixed sprint budget

An exit arrangement your customers can test

For cloud services, leaving must not be made unnecessarily difficult, and switching costs are being phased out. We build the export route: which format, within what time, how complete, and with the option for the customer to try it out in advance. An export that only proves unusable at the point of departure is not an exit arrangement but a formality, and you only find out when it is too late to do anything about it.

Full exportTestable in advanceTransition periodDocumentation

What you have at the end of an engagement.

Working integrations and a permissions model, plus the documentation you need when a customer or a regulator asks how it is set up.

  • A separation between raw and derived dataMeasurements and events in a layer you can open up, and your own analyses and insights in a layer that stays yours.
  • An integration for accessA documented route through which the user, and third parties they designate, can request data in a common, machine-readable format.
  • A permissions layer with revocationPermissions per party and per type of data, with a duration, and a button that lets the user end shared access themselves.
  • An access logWho requested what and when, retained in line with your retention policy, so that access can be audited afterwards.
  • A testable export routeA full export in a common format that your customer can try out in advance, with documentation on the structure and meaning of the fields.
  • Handover to your teamWorking sessions with your developers and with the people who will handle the requests, plus the documentation your lawyer needs for the contractual side.

When this applies to you.

Four situations we see in organisations that need to get to grips with this. There is usually a concrete trigger, and the scope often turns out to be broader.

Connected products

Your devices send data to your platform

You supply machines, vehicles, meters or equipment that send measurement data to your own environment. From 12 September 2026, newly marketed products must be designed so that this data is accessible to the user by default.

Revenue model built on exclusivity

Your service relies on you being the only one who sees the data

Maintenance contracts, usage advice or optimisation that you can offer because you alone hold the measurement data. That foundation is shifting, and it is wise to decide now where your value lies afterwards.

Cloud service

You provide a service that customers live in

The regulation requires that leaving is possible without unnecessary barriers and phases out switching costs. Quite apart from the obligation, this is a selling point: those who can show how easy it is to leave win out over those who are vague about it.

Requests are already coming in

Data has been requested and nobody knows how to handle it

A customer, or a party they have engaged, has asked for access and the answer was a manual export. That is not sustainable once there are more of them, and it is precisely the pattern the regulation aims to eliminate.

How a project works.

1

Introduction and scoping

Which products or services it affects, who the user is in your case, and what data the product actually produces. That last question sparks more debate than expected, because far more is usually recorded than is ever used.

2

The line between raw and derived

A working session in which we determine, for each type of data, whether it falls under what the product generates or what you have made of it. The regulation does not draw this line sharply, so we document the reasoning. That is what you can explain later, even if the interpretation shifts with practice.

3

Building in sprints

We work in sprints and start with access for the user themselves, because that is the simplest and everything else builds on it. Then the permissions layer for third parties, then the export route. Your own developers work alongside us.

4

Testing with a real request

We walk through a complete request with someone from your organisation in the role of user: requesting, sharing with a third party, and revoking. We also have an export actually imported somewhere else. Only then do you know whether the route works.

5

Documentation and handover

The documentation for your customers and for your lawyer, a handover to the people who handle the requests, and agreements on what happens when a request comes in that falls outside the beaten track.

Frequently asked questions about the Data Act.

What product owners, architects and directors ask us before they get started.

When does this apply?
The regulation entered into force on 11 January 2024 and has applied since 12 September 2025. For connected products placed on the market from 12 September 2026, a design requirement also applies: the product must be built so that the data is accessible to the user by default and with ease. The provisions on switching charges for cloud services follow their own timeline, under which those charges are phased out gradually.
What exactly counts as "data the product generates"?
Raw data and what follows directly from it: measured values, states, events, usage data. What falls outside is information that results from substantial investment in processing and analysis. That line is not drawn sharply, and there will be debate about it in the coming years. For your design, it mainly means you want to keep the two separate, so you can open up one layer without pulling your system apart.
Can we charge for access?
For the user requesting their own product data, access must not be made unnecessarily difficult and there are restrictions on what you may charge. When making data available to a third party at the user's request, different rules apply, including the possibility of reasonable compensation, with a lighter regime for small businesses. This is above all a question for your lawyer; we make sure your system can do both.
What happens to our trade secrets?
They remain protected. The regulation expressly provides a mechanism that lets you attach measures to the sharing of data containing trade secrets and, in exceptional cases, refuse to share it. What does not work is declaring everything a trade secret to avoid the obligation. In practice, separating raw from derived data largely solves this: your secret lies in your analysis, not in the measured value.
Do we have to provide this in real time?
Where the product generates the data continuously or almost continuously and this is technically feasible, immediate availability is the obvious choice. For many applications, periodic delivery is enough. What matters most is that access is simple, secure and in a common format, with no one on your side having to step in. We design the integration so you can adjust the rhythm later without rebuilding.
How long do we have to keep data available?
That can't be captured in a single retention period. It depends on the nature of the product, on what you agree contractually and on other retention obligations that apply to you. What we do is substantiate and document that choice rather than guess at it, and set up the retention policy so that it differs per type of data. That saves on storage costs and makes it easier to explain.
We are a recipient, not a supplier. Is this useful to us?
Absolutely. As a recipient you gain rights: access to the data from the equipment you use, and the ability to have another party work with it. That opens up maintenance contracts that were previously locked in and makes comparison possible. On that side, we help with retrieving that data and making it usable, and with the question of what you need to put in your contracts to actually receive it.
How does this relate to the GDPR?
They apply side by side. If the product data concerns an identifiable person, the GDPR applies in full and you need a legal basis for it; the Data Act does not loosen that. In practice, most machine data is not personal, but with vehicles, wearables and equipment in a home that is different. We flag this in the workshop so that your legal counsel knows what to look at.
What does this mean for our architecture?
Three things. Data must live somewhere it can be accessed without anyone having to go into your system; that means an integration, not database access. There must be a permissions layer around it, because giving access to the user is not the same as giving access to everyone. And there must be logging, so that access can be audited. These three are manageable work if you include them in the design, and a rebuild if you have to add them later.
What belongs in an exit arrangement?
The format in which data is handed over, within what time, under what conditions, and how long you keep supplying after termination so that the transition can be completed. That last point is the one most often missing and the one that gives the most peace of mind. Add to this that the customer may test the export format in advance – that is the difference between an arrangement that works and one that only exists on paper.
Will this cost us our customers?
That is the fear, and it is rarely the outcome. Customers mainly leave when they feel they cannot leave. In procurement processes, a demonstrably working exit arrangement is now a plus point, not a risk. What you do need to think about is which part of your service relied on exclusive access – that part becomes vulnerable, and that is a strategic question better raised now than later.

Talk to us about the Data Act.

A no-obligation introductory call of half an hour. Tell us what your product produces and who uses it, and we'll tell you where your architecture is likely to be under strain and what needs to come first – even if we end up not working together.

Edit content