MVP and SaaS
From MVP to SaaS: when should you migrate, and what will it cost?
Lovable, Bolt, Bubble or Base44: the signs an MVP has run its course, what you can keep, and what it costs to move to a production-ready application.
The short answer
An MVP built with Lovable, Bolt, Bubble or Base44 needs to evolve when its limits hold back usage: regressions, poorly protected data, impossible integrations or rising costs. Migrating does not mean rewriting everything: an audit decides, journey by journey, what is kept, migrated or rebuilt. At CUB3, a limited-scope migration costs €3,600 to €12,000 excluding VAT, audit included.
- Migrate when the MVP’s limits hold back usage or sales, not because the tool is no-code.
- What you can keep depends first on what can be exported: the code, the database or only the data.
- At CUB3, a limited-scope migration costs €3,600 to €12,000 excluding VAT, audit included, with the budget confirmed after the audit.
- UnderstandCode, access and journeys
- SecureBackups and version control
- RelaunchUseful features and tests
Signs an MVP has run its course
An MVP exists to test whether a need is real, with as little effort as possible. Tools such as Lovable, Bolt, Bubble or Base44 do this very well. The difficulty comes later, once the product has users: you then have to evolve it without breaking what works, protect real data and answer customers who ask about reliability.
The tool used is not enough to decide. A no-code MVP can remain suitable for a long time, and AI-generated code can be taken over. It is the signs below that show a change of approach is needed, especially when several of them add up.
- Every change causes regressions in journeys that used to work.
- Nobody can say precisely who has access to which data.
- An integration a customer has asked for is impossible or very fragile in the current tool.
- The platform’s cost is rising faster than usage or revenue.
- A customer, partner or investor asks for guarantees you cannot provide.
- New developments take longer and longer, even for small changes.
What you can keep depends on what you can export
The first question is practical: what can you get back? As of November 2026, Lovable and Bolt let you retrieve the application’s code. Base44 lets you export the frontend code, but its database and server-side logic remain hosted on its platform. Bubble does not let you export code: only the data can be exported.
Getting the code back does not mean keeping it as it is. Quickly generated code may work without tests, with data access rules that need reviewing. Conversely, Supabase can be kept when the way it is used meets your needs: switching services should address an identified constraint, not a preference.
| Starting point | Often reusable | Often to rework or rebuild |
|---|---|---|
| Exportable code (Lovable, Bolt) | The code, and the Supabase database if the project is in your name | Tests, data access rules, parts that break with every change |
| Exportable frontend, hosted backend (Base44) | The frontend code and exported data | The database and server-side logic |
| Platform without code export (Bubble) | Exported data, validated journeys and business rules | The application, rebuilt around the selected features |
| In every case | User feedback and the list of journeys actually used | Deployment, backups and the ability to roll back |
Keep, migrate or rebuild: decide journey by journey
A useful migration does not carry everything over. It focuses on the essential journeys: login, creating a request, viewing a record, payment or administration, depending on your product. Other features wait, or are dropped if nobody uses them. A new feature does not belong in the migration just because it would be useful for the next version: it is priced separately.
For each selected journey, the audit chooses between three outcomes. A data migration and a functional rebuild involve neither the same work nor the same budget: that is why the price is confirmed after the audit, not before.
Keep
What exists works, can be tested and does not block planned developments.
Migrate
The logic has been validated by real use, but the tool is a constraint on reliability, integrations or costs.
Rebuild
The journey breaks regularly, cannot be exported or does not protect data properly.
The budget: €3,600 to €12,000 excluding VAT for a limited scope
At CUB3, moving from an MVP to an adaptable application costs €3,600 to €12,000 excluding VAT for a limited scope. This budget covers the audit of the existing product, migration of the selected data, migration or rebuilding of the defined features, journey testing and cutover. It is confirmed after the audit, which is not charged a second time.
A complete redesign, complex features or advanced scaling require a separate quote. Hosting, domains, licences and third-party services remain at your expense. A full staging and production infrastructure, with Terraform and continuous deployment, falls under the DevOps offer, at €6,000 excluding VAT.
| Item | Coverage |
|---|---|
| Audit of the existing product: features, data and dependencies | Included in the €3,600 to €12,000 |
| Migration of selected data | Included, depending on available exports |
| Migration or rebuild of selected features | Included, within the defined scope |
| Journey testing and cutover | Included |
| Complete redesign, complex features, advanced scaling | Separate quote |
| Staging and production infrastructure with Terraform | DevOps offer, €6,000 |
| Hosting, domains, licences and third-party services | At your expense |
Prepare the cutover
If your MVP already has users, the switch to the new version needs to be prepared as an operation in its own right. You need to know which data is migrated, in what order, and what happens if a check fails. A zero-downtime cutover cannot be taken for granted: it depends on the product, the data and the access available.
Also define what “production-ready” means for you before you start. These are verifiable criteria, agreed during scoping: journeys to test, data to check, access rights and deployment conditions. An advanced load or availability requirement needs a separate assessment.
- A full data backup exists before the cutover.
- A rollback to the previous version is prepared.
- Users are notified of any interruptions.
- Migrated data is checked after the cutover.
- Essential journeys are tested in real conditions.
HODNOS: keep little, rebuild what matters
HODNOS was not a no-code MVP, but the decision was the same. The platform connects customers with artists. At takeover, booking, payment and messaging no longer really worked, and the PHP code was edited directly on the server, without version control.
CUB3 first recovered the code and created the Git repository, then identified the few reusable elements. The team then rebuilt the essential journeys, from registration to back-office, moved hosting to AWS with Terraform and delivered web, iOS and Android. The lesson applies to an MVP: establish the facts before choosing what to keep.
What to gather before the audit
The audit goes faster, and the budget can be confirmed more reliably, when access and information are ready. Provide access to the tool used and its exports, list the journeys actually used and describe the problems you have observed. Usage figures, even approximate ones, help determine what must hold up in production.
- Admin access to the MVP tool and, if possible, to the code or repository.
- Available data exports and the list of connected services.
- The journeys your users actually rely on.
- Observed problems, with examples.
- The approximate number of users and the platform’s current cost.
In the field
A real project
A platform without version control and with broken journeys: recover what exists, keep what is useful and rebuild web, iOS and Android.
The HODNOS recovery