A Right to Switch Is Not an Exit Plan

Most technology contracts are signed with a clear view of the way in and a vague view of the way out. The onboarding plan has milestones, owners and a date. The exit plan, if it exists, is usually a clause: data will be returned “in a commonly used format” within a stated period. It reads as reassuring, but nobody has tested it.

European regulation is now changing the legal side of that picture. That makes it a good moment to ask a harder, more practical question. If you had to leave your main supplier next quarter, could you?

What the regulator says

The EU Data Act has applied since September 2025. In the Netherlands, the Authority for Consumers and Markets (ACM) supervises it. On 19 October 2026 the ACM published guidance on switching between cloud services. Its stated principle is that switching should be quick, easy and free of charge.

The guidance groups provider obligations into four themes:

• No obstruction. Providers may not obstruct switching or parallel use, and must cooperate while maintaining service continuity.
• Commercial terms. Customers receive clear information in advance, and contracts record the switching arrangements.
• Technical support. Providers offer free APIs for data exchange and work with interoperability standards. For infrastructure services (IaaS), there is a requirement of functional equivalence after the switch.
• Protection from foreign government access. Providers protect non-personal data against unlawful access or transfer by governments of countries outside the EU.

On cost, the ACM says that from 12 January 2027 providers may no longer charge for switching. Until then, they may charge only costs directly linked to the switching process, such as data transfer and technical support. For parallel use, only the actual cost of data transfer may be passed on, including after January 2027.

The ACM is also careful about what the rules do not do. It says switching remains a far-reaching process that involves organisational costs. What the Data Act removes is uncertainty and costs imposed by the provider. That distinction is the subject of this article. The regulation addresses the supplier’s behaviour. It does not address the customer’s readiness.

This is a summary of regulator guidance for business readers, not legal advice. Whether and how the Data Act applies to a specific contract is a question for qualified counsel.

What businesses report

The ACM has also asked users what happens in practice. In 2025 it ran a quantitative questionnaire among 420 business users of cloud services, mainly working in small and medium-sized businesses. Two findings matter here:

• 61.7% of respondents said they had never switched cloud provider.
• Of those who had attempted to switch, 30.2% did not succeed. That figure applies to people who tried, not to all 420 respondents.

These answers are self-reported and come from a single national sample. They do not explain why attempts failed. The cause could be technical, contractual, financial, or a decision to stop partway. They also come from before the guidance was published. So they show the starting position, not whether the new rules work. They are still a useful corrective to the idea that leaving is a routine administrative step.

Our analysis: the gap between a right and a capability

A switching right tells you what a supplier must allow. An exit plan tells you what your organisation must be able to do. The two are easy to confuse, because both appear in the same contract conversation.

In our view, most of the effort in a supplier exit sits on the customer’s side. Someone has to know where the data lives and what depends on it. Someone has to rebuild integrations, move identities and permissions, and retrain staff. Someone has to keep the service running while all of this happens. A free export API does not do any of that. A data file that is complete but has never been restored somewhere else proves very little.

This also changes what a renewal decision really is. It is rarely a binary choice between staying and leaving. A buyer usually has five options:

1. Renew on current terms, knowingly.
2. Renegotiate, using a credible alternative as leverage.
3. Migrate to another provider.
4. Run in parallel for a defined period, which the Data Act explicitly contemplates.
5. Stop the service or capability altogether.

Each option is only as credible as the evidence behind it. If you cannot show that migration is feasible, you cannot really use it as leverage when you renegotiate. Treated this way, an exit plan is not pessimism about the supplier. It is what makes every other option real.

A six-part exit acceptance check

Here is a practical test to run before signing or renewing. It is a set of questions to answer with evidence, not a standard or certification.

1. Data export and restoration. Can you export a complete copy of the data, including metadata, history and attachments, in a usable format? More importantly, has anyone restored it into another environment and checked it? An export that has never been restored is an assumption.
2. Dependencies and integrations. Which systems, automations, reports and partners connect to this service? Who owns each connection, and what breaks if the service disappears for a day? Hidden dependencies are where exits stall.
3. Security and identity. How are users, roles, keys and audit logs moved, rebuilt or retired? Can access be cut off cleanly at the old provider without leaving orphaned accounts or credentials?
4. Service continuity and recovery targets. What is the acceptable downtime (recovery time objective, RTO)? What is the acceptable data loss (recovery point objective, RPO)? Can the exit realistically meet those targets, or does it need a parallel run?
5. Contract, legal and cost. What do the notice periods, termination fees before January 2027, data-retention commitments and transfer costs actually say? What is the realistic internal cost in people and time? The regulation can remove provider-imposed charges. It cannot remove your own project costs.
6. Owner and rollback. Who is accountable for the exit decision and its execution? What is the point of no return, and what happens if the move has to be reversed halfway through?

If you can’t answer most of these questions, it doesn’t mean you should leave. It means you don’t yet know whether you could.

Fair counterarguments

There are reasonable objections to this approach.

“Exit planning is wasted effort if we will never leave.” Sometimes that is true. For a low-risk, easily replaced tool, a light check is enough. But how much effort to put in should depend on how badly the business would be hurt if it had to leave, not on how likely leaving seems today.

“The regulation solves this.” It addresses part of it. Supplier obstruction and many provider charges are now in scope, but the ACM itself says organisational costs remain.

“Multi-cloud is the real answer.” Using several providers can reduce dependence on one, but it adds complexity, cost and its own integration risks. It is a design choice with trade-offs, not a substitute for knowing how you would leave any single provider. It does not automatically make a service more resilient.

“Our supplier is a trusted partner.” That may well be so. An exit plan is not a sign of distrust. It gives the relationship a clear, tested boundary, and good suppliers tend to welcome that clarity.

What this means for the buying decision

Some open questions remain. We don’t know whether the new guidance will measurably reduce failed switches. Its effect on SME negotiating power is also unproven. These are hypotheses worth watching, not conclusions.

The practical point does not depend on the answers. Before the next renewal, ask whether you know how you would leave, and whether you have tested any part of it. If the answer is no, the right next step may be modest: a restoration test, a dependency map, or a named owner. It may also be that the current supplier remains the best choice once the alternatives are honestly compared.

If you want to work out which of the five options is realistic for your situation, the free Workflow Fit Diagnostic is a structured place to start, and it may be all you need. — Ronny Boesing