There is a widespread assumption that the move to SaaS has made data migration a fading concern. The logic seems reasonable: if the vendor runs the platform and there are no more big-bang upgrades, surely the era of painful migrations is over.
It is a myth, and it is an expensive one.
What SaaS genuinely removes is a category of operational pain: self-managed upgrades, patching, and version drift. That is real value. But it says nothing about what happens when you change providers, consolidate systems after a merger, or redesign a core process. Each of those is a migration, and SaaS does not make the data move itself.
The difficulty simply relocates:
- Schema differences. Your current provider’s data model and the target’s are not the same, and the gap between them is exactly the mapping work migration has always required.
- Process coupling. Business processes are wired to the current configuration. Moving the data without understanding that coupling moves the problems with it.
- Extraction reality. Getting your data out of a SaaS platform in a complete, usable form is frequently harder than expected, and the contract terms around it matter.
- Ownership boundaries. The line between what the vendor is responsible for and what you are responsible for is where the Responsibility Gap reappears, often unnoticed until something fails.
From a business perspective, the danger is precisely that the work is underestimated. Because the platform feels managed, leaders assume the data will “just come across”, budgets are set too low, and timelines do not allow for discovery and reconciliation. The result is the same late-stage surprise that on-premise migrations were notorious for, only now it arrives wearing a SaaS badge.
The PDM response is not different in kind, which is the point. The same discipline applies: understand the landscape before committing to a date, make ownership explicit across the client and supplier boundary, express quality expectations as testable rules, and treat extraction and mapping as real engineering, not a download.
SaaS changed where the battle is fought. It did not end the war. Organisations that recognise a provider switch as a genuine migration, and resource it accordingly, keep their optionality. Those that believe the myth discover the cost at the worst possible moment.
Where have you seen a SaaS switch treated as a simple export, only to become a full migration once the data model and process coupling showed up?
Compare your next planned SaaS move against one PDM module, and pressure-test the assumption that the data will just come across.