The challenge
The Boomerang was running on Bubble.io. That is a perfectly good place for a product to start, and we spent roughly six to eight months building and extending the Bubble application with the team. No-code gets an idea into the hands of real users faster than anything else, and for a while it was the right call.
What surrounds a Bubble build, though, tends to grow quietly. Alongside the application the team was also paying for Airtable, which had become the working surface for their data team — a second subscription, a second system, and a second place where the truth about the business lived. Every operational process ran across both.
Eventually the team hit the point every successful no-code product reaches: the ceiling of the platform started to cost more than it saved. They decided to move to a custom solution, and asked us to build it — the same team that had built the Bubble version, which meant nobody had to relearn the domain from scratch.
The Bubble.io engagement
We worked on the Bubble.io application for around six to eight months before any rebuild was on the table. That period is the reason the replatform went the way it did: we were not handed a specification, we already knew how the product actually behaved, which workflows mattered, where the edge cases were and which features were load-bearing.
It also meant we could be honest about when Bubble stopped being the right answer. We do not push clients off no-code as a matter of policy — several of the products in this portfolio are Bubble builds we still recommend keeping there. The Boomerang moved because the team decided a custom stack served them better, and we agreed.
A custom rebuild on MERN, Next.js and Supabase
We rebuilt the product as a full custom application: a MERN-stack application layer — React on the front end, Node.js and Express services behind it — with Next.js delivering the site, and Supabase as the database.
Next.js gave the public-facing side proper rendering control, routing and performance characteristics that a no-code platform simply does not expose. Supabase gave the team a real relational database they own, with direct access to their own data instead of it living inside a platform's abstraction.
The rebuild covered the whole site and application rather than a facade over the old system, so there is no residual dependency on the previous platform holding the architecture back.
Data Team Panel, n8n automations and Intercom
The single most useful piece of the rebuild was the Data Team Panel. Their data team had been working in Airtable — a paid tool sitting outside the product, holding data the product needed. We built a custom panel inside the application to do that job, so the data team works against the real database in the real system, and the Airtable dependency went away with it.
We built complex n8n automations to carry the operational workflows that connect the moving parts of the business — the sequences that had previously depended on manual steps or on the Bubble platform's own workflow engine.
We also went through their Intercom setup and resolved a long list of issues in it. Support tooling is usually the last thing anyone budgets attention for, and it is where a surprising amount of daily friction accumulates.
The outcome
The Boomerang now runs on a stack it owns end to end: a custom MERN and Next.js application on Supabase, with the data team working inside the product rather than in a separate paid tool, and operational workflows handled by n8n automations we built for them.
The Airtable subscription that had been running alongside the Bubble application is gone, replaced by an in-product panel — one system instead of two, and one place where the data lives.
For us it is the clearest demonstration of something we say often: the right stack changes over the life of a product. We built their no-code version, we told them honestly when it had served its purpose, and we built the custom replacement. It is the case study behind our MERN, Next.js, replatforming and n8n automation work.