Bubble to production: keep the workflows, stop treating editor Run as the operating model
Bubble is a no-code application platform: you draw pages, bind data types, write privacy rules, chain client and backend workflows, and deploy a Live version onto Bubble hosting. That loop is excellent for proving a journey. It is the wrong finish line once customers need rules that reject another user’s unique id, a Workload Unit budget with a kill switch, and a deploy that is not “whoever has the editor tab.” Production starts when signup through one write survives on the custom domain with the editor closed.
CodeCross LLC describes bubble-to-production as the moment a Bubble app leaves the editor’s Run as proof: privacy rules reject another user’s unique id, Live is distinct from Development, and Workload Units have a budget with a kill switch. Proof is signup through one write on the custom domain with the editor closed. If only a Bubble admin can ship Live, the app is not operated.
What Bubble preview hides
A green editor Run is not an app you can operate
Bubble compresses page-to-Live into one surface. Real users expose privacy rules that fail open, workflows that burn Workload Units, and a custom domain that still depends on someone opening the editor to deploy Live.
Privacy rules only “worked” because you were the Current User
A search that returns another user’s things by unique id is a production incident. Rules belong on the data type, not on a hidden button.
API Connector and plugin keys live only in the editor
Collaborators rotate tokens by opening the plugin pane. A second engineer cannot inject destination secrets from a clean seat, so on-call becomes “who has Bubble admin.”
Rollback means undeploy Live or restore a version by hand
If the only reverse is “open Bubble and deploy an older version,” you do not have a release. Live versus Development plus a named prior version must be rehearsed.
Closing the editor stops the next Live ship
Availability that requires Bubble’s editor to mutate workflows is still a lab. Production stays up when the project tab is closed.
The editor-to-Live pass we actually run
Review privacy rules, then cut over the advertised domain
Keep the pages and workflows that already convert. Move Live authority to seats that can restore a version and rotate plugin keys without opening every tab. Bubble remains a workflow workshop until that domain is proven.
01
Write privacy rules that reject a guessed unique id
Every load-bearing data type gets a rule a second operator can read. Current User convenience searches come off the critical path.
02
Split Development from Live; inventory API Connector keys
Mint fresh values for Live. Development credentials stay out of the advertised domain. Plugin panes stop being the production vault.
03
Cap Workload Units and name a kill switch
Recursive workflows, recursive searches, and public Data API reads get budgets. Alert before a stranger can invoice you from the public app.
04
Prove signup → session → write with the editor closed
Logs tag the Live version on the custom domain. If those hold, the app shape stays; the control plane has left editor Run.
How to
Move a Bubble app onto a Live version you can operate
Privacy rules, Live/Development split, and Workload Unit budget. Success is signup and one write after you close the editor.
Step 01
Treat editor Run as journey evidence, not Live proof
Keep the Bubble-shaped pages if navigation still matches what operators will pay for. Do not treat a green preview as durable ownership of rules or keys.
Step 02
Review privacy rules against another user’s unique id
Each mutating data type must deny a guessed id. Document which searches are Current User convenience versus server-enforced.
Step 03
Split Development from Live and remap API Connector names
Mint new secret values. Split preview from production. Never leave Live keys only in the plugin pane.
Step 04
Set a Workload Unit budget and a kill switch
Cap recursive workflows and public Data API. Point the custom domain at Live only after those caps exist.
Step 05
Run the core journey with the editor closed
Signup, session, and one write on the custom domain. Confirm logs tag the Live version. Editor Run is then optional for availability.
Read next
Proof, the essay, and sibling intents
These pages are already on the site. Use them to pressure-test the bet before a call.
Why is a Bubble editor Run not enough for production readiness?
Editor Run proves pages render and a happy-path workflow returns. It does not prove privacy rules reject another user’s unique id, Workload Units have a budget, or a teammate can deploy Live without the person who built the first page. If keys, rules, and deploys still live only inside the editor, you have a convenient no-code path—not an operated app.
What must be true before you cut over a custom domain on Bubble?
Privacy rules on every load-bearing data type must deny a guessed unique id. Development and Live must use different API Connector and plugin values. Workload Units need a budget and a kill switch. Document which workflows are client versus backend so the next operator does not reverse-engineer the app from a page.
When should Bubble Live stay versus when should a destination own the runtime?
Keep Bubble Live while the app is still a workflow-shaped MVP and editor risk is acceptable. Move the runtime when source export, second-engineer deploys, or Workload Unit cost become the risk. Write down which model production uses so on-call does not discover “only deploy Live in Bubble works” during a launch spike.
How do you prove production without redrawing every Bubble page?
Run the core journey on the custom domain with Live credentials. Confirm privacy rules fail closed, Workload Units stay under budget, logs tag the Live version, and closing the editor does not stop the advertised domain. If those hold, the app shape stays; the control plane has matured past editor Run.
Make editor Run optional for availability.
Bring the data-type list and the custom-domain story. We will say whether this week is an operated Live version — or another preview pretending to be production.
“What impressed us most about CodeCross was their ability to deeply understand our vision and translate it into a complete digital solution. Unlike many agencies that just focus on technical delivery, CodeCross approached our project like true partners.”