Vendor lock-in
In shortNot the impossibility of leaving, but its price. That price can be measured.
Lock-in is usually described as a condition: trapped, dependent, at somebody's mercy. It is more useful to treat it as a number. What would it cost to switch within the next twelve months — in person-days, in consultancy fees, in downtime?
You can estimate that number. And once you have it, it changes negotiations.
The four levers that set the price
Data format. An export into a proprietary format is not an export. Check whether the data comes out in a form another tool can read without custom development.
Interfaces. Every application written against a vendor-specific API is its own migration project. Standards such as S3, SAML, OIDC, CalDAV or ActivityPub cut this item dramatically.
Process knowledge. The most expensive part is rarely the technology. It is the 400 automations, forms and reports that grew inside a system over years and that nobody has fully documented.
Contract commitment. Multi-year agreements with volume discounts are the cheapest lock-in instrument a vendor has: leaving is not forbidden, merely made unattractive.
What the Data Act changes
The switching provisions of the EU Data Act have applied since September 2025. Cloud providers must support a switch, switching charges have been withdrawn, and functional equivalence is mandated for infrastructure services. See EU Data Act.
That lowers the bill; it does not remove it. No regulation covers process knowledge.
Handling it in practice
Lock-in is not always wrong. A deeply integrated system that works well is often worth more than the theoretical freedom to leave it.
What is wrong is not knowing the price. Knowing it means deciding deliberately — and negotiating differently, because both sides know a switch is within reach. See Exit strategy.