The buyer of a workbook holds something permanent. The subscriber to a platform does not.

What ownership means here

The file keeps working if the company does not.

An owned workbook has an unusual property for software: its continued function depends on nothing except the operator's machine. If the company ceased trading tomorrow, every calculation would still run, every record would still open, and the governance layer would go on refusing what it refuses. There would be no more updates and no support, and the instrument itself would be unaffected. That is a genuinely strong position for a buyer and it is not the norm in this category.

The dependency a service introduces

Continuity becomes something the operator does not control.

A hosted edition adds a party to the arrangement whose continued existence and continued willingness both matter. Service can end because a company fails, because a product is discontinued, because terms change, or simply because a payment lapses at an inconvenient moment. None of these is likely in any given month and all of them are outside the operator's hands. For most software that is an ordinary inconvenience. For the layer that governs deployment, it is a dependency worth naming out loud, because the whole discipline being sold is about not carrying uncontrolled risks silently.

What the service buys in exchange

The rented version can improve continuously; the owned one cannot.

The compensating advantage is genuine. An owned artefact is frozen at purchase and improves only when the operator obtains and installs a new one, which in practice means most owners run an older edition than the current one. A service is always current, which matters most for the parts of the system that should be revisited as evidence accumulates. It also carries costs the buyer never sees — hosting, maintenance, security — which are what the recurring price is actually for.

The commitment that follows

Records must be exportable in a form that outlives the service.

The obligation this analysis creates is concrete rather than reassuring. If a hosted edition holds an operator's history, that history has to be retrievable in a format that remains usable after any relationship with the company ends — not a proprietary archive, but records the operator could analyse independently. That converts the worst case from losing the evidence base to losing the tooling around it, which is a materially different kind of loss and the only version of the risk that is acceptable.

• Losing the tooling is survivable. Losing the record is not.
• Export has to be usable without the product that produced it.
• The obligation belongs in the design, not in a support policy.

Why both are being offered

The two models suit different buyers, and neither supersedes the other.

The owned edition is not a legacy step on the way to the real product. An operator who values permanence over currency, who is uncomfortable with a recurring dependency, or who simply wants the instrument to be a thing they possess is well served by it, and will remain so. The hosted edition suits an operator who wants the machinery to keep improving without managing it. Framing the second as an upgrade would obscure a real trade that different buyers should make differently.

The key idea

Name what the buyer stops holding, especially when the change looks like a benefit.

A move to a service is usually presented entirely in terms of what arrives — better interface, continuous improvement, nothing to maintain. Something also leaves, and in this case what leaves is the guarantee that the instrument continues to function irrespective of the company that made it. That is worth stating in the same paragraph as the benefits, by a product whose central claim is that unstated risks are the expensive ones.