Bubble vs custom code: when to switch, and how to move
Stay on Bubble while it is cheap and fast to change. Switch when workload costs, speed, logic or due diligence start to hurt. The signs and the steps.
By Sohail Hussain21 November 20266 min read
Stay on Bubble while it is cheap to run and fast to change, and switch to custom code when one of five things starts to hurt: workload costs rising faster than revenue, pages slowing as data grows, logic that has become too complex to change safely, a feature the platform cannot do, or an investor or enterprise customer asking what you own. One of these is a reason to plan. Two or more is usually a reason to start.
Bubble is a good place to begin a product. It is rarely the right place to finish one that has grown. The trick is to leave at the moment the rebuild is affordable and the product is well understood, not in a crisis. We rebuild Bubble apps in code as part of our web app and SaaS development work, and we also tell founders when staying is cheaper.
What does each option give you?
| Bubble | Custom code (for example Next.js and PostgreSQL) | |
|---|---|---|
| Speed of change | Very fast for a founder who knows the editor | Fast with a team; slower for one person |
| Running cost | Plan price plus workload units, rising with usage | Hosting and services, usually flatter as usage grows |
| Performance | Fine for small data; slows with large searches and lists | Tunable with indexes, caching and background jobs |
| Testing | Manual, mostly | Automated tests on the paths that matter |
| Version control | Bubble's own branches and save points | Git, code review, rollbacks |
| Ownership | Your data; the app runs only on Bubble | Code, data and hosting accounts are yours |
| Hiring | Bubble specialists | Any developer in the stack |
The ownership line comes from Bubble itself. Its application and data ownership page says there is no way to export an app as code, and that moving off the platform means rebuilding the logic. That is not a reason to avoid Bubble. It is a reason to know your exit cost before you need it.
What does Bubble cost as you grow?
As of October 2026, Bubble's pricing page lists these plans, priced per app, billed annually:
| Plan | Price per month | Workload units per month | Notes |
|---|---|---|---|
| Free | $0 | 50,000 | For building and testing, not for a live app |
| Starter | $59 | 175,000 | Custom domain, live app |
| Growth | $209 | 250,000 | 2 editors |
| Team | $549 | 500,000 | 5 editors |
| Enterprise | Custom | Custom | Custom server options |
All plans cover web and native mobile on one project. Monthly billing costs more than the annual prices above.
Workload units measure the server work your app does: searches, workflows, API calls, page loads. Usage grows with users and with how efficiently the app was built. An app with heavy searches on every page can use several times the workload of a leaner one with the same users. When the bill starts climbing, the first fix is often optimising searches and workflows, not leaving.
The signal to watch is the ratio. If the platform cost is a few percent of revenue, it is cheap. If it is heading towards a large share of gross margin, and optimisation has been tried, a rebuild starts to pay for itself.
What are the signs it is time to switch?
Workload costs are outgrowing revenue. You have optimised the expensive searches and workflows, and the bill still rises faster than income.
Pages are slow and users notice. Large lists, complex searches and nested data slow down as the database grows. Support tickets mention speed.
Changes have become risky. Nobody is sure what a workflow change will affect, bugs reappear, and the founder is afraid to touch the core flow. In code, this is the point where you add tests. In Bubble, the options are limited.
You need something the platform cannot do well. Heavy AI processing, long-running jobs, complex integrations, fine control over data location, or performance guarantees for a large customer.
Someone is asking what you own. Investors in due diligence and enterprise buyers in security reviews ask about code ownership, hosting regions and disaster recovery. A Bubble app can pass some reviews and not others.
If only the first sign applies, optimise first. If two or more apply, start planning the move, while the product still earns enough to fund a rebuild.
How does a move from Bubble work?
1. Write down what the app actually does
Go through every page, workflow, backend workflow, plugin and API connection, and list what each does. Then check usage: features nobody uses are not rebuilt. This step usually cuts a fifth or more of the app.
2. Map the data
Bubble's data types become database tables. Enable the Data API in the app's API settings and pull each type as JSON, or export CSV files from the data editor. Paste a sample into our JSON to TypeScript tool to get typed definitions for your migration scripts. Watch for Bubble's habits that do not translate directly: lists of things stored on a record, option sets, and fields that hold references to other types. Each needs a decision about tables and relationships.
3. Plan how users will sign in
Bubble's data export does not include users' passwords. Decide early how users get into the new system: a "set your password" email before cut-over, a magic link on first sign-in, or Google sign-in for those who used it. Tell users before the move.
4. Rebuild the core workflow first
Build and test the workflow customers pay for before anything else, then the rest in order of use. Run test imports of real data throughout, not just at the end.
5. Cut over in one move
Pick a quiet period, put the Bubble app into read-only mode, run the final data import, check counts and spot-check records, then switch the domain. Keep the Bubble app on its plan, read-only, for a month in case something was missed.
When should you stay on Bubble?
More often than founders expect.
- Revenue is small or unproven. A rebuild spends months of runway on a product that may still change direction. Stay until the product is stable.
- The app is an internal tool. Fifty staff using an approvals tool will not notice the difference between Bubble and custom code.
- The founder is the builder. If the founder changes the product daily in the editor and that speed is the advantage, a rebuild slows the company down for months.
- The problems are fixable in place. Slow pages from poorly built searches, and high workload from wasteful workflows, can be fixed by a Bubble specialist for far less than a rebuild.
A middle path is to move only the heavy parts. Put complex logic, large data or AI work behind an API you own, and let Bubble call it. The Bubble app stays the interface, workload drops, and the new backend becomes the start of the eventual rebuild.
No-code vs custom development covers the decision from the other side, for founders choosing where to start. The full path from idea to paying customers, including where a no-code first version fits, is in our guide on how to build a SaaS product.
Was this guide useful?