Captain's log, stardate d308.y42/AB

A good share of the projects we take on through our Rescue Squad start the same way. A company is unhappy with its current development provider and decides to switch. Missed deadlines, poor quality, a team that stopped answering, or, like in the case of Naiz, their provider shut down overnight. They call us to take over and they need us to do it right now.
Most of the time, the handover is uncomfortable but manageable. Sometimes it isn't.
We recently started working with a company that is in an active legal dispute with its previous provider. The provider controls the repositories, the servers and the accounts where the company's data lives, and it's not making the migration easy. Meanwhile, the company needs that access to operate every day. Every week without it is a week of operational damage, regardless of who ends up winning the dispute.
It's not the first time we've seen this. And in almost every case, it could have been prevented before the first line of code was written.
When a company hires a development provider, it's buying expertise, time and execution. It's not buying a license to use its own product.
That's how we frame it in our contracts at MarsBased. The client hires our experience, our hours and our work to build an application. The source code and the intellectual property of what we build belong to the client from the signature of the contract. Most other companies do it upon receiving the final payment but we don’t like that approach. We give IP and the code right away so we avoid falling into legal disputes later on, if things go south.
On the operational side, we've worked both ways. Sometimes directly on accounts owned by the client. Sometimes on MarsBased accounts that we transfer to the client when the project ends, or whenever they ask. Both work, as long as ownership is clear on paper and the transfer is a formality, not a negotiation.
That's the point of this article. Ownership has to be real at two levels: legally and in practice.
These are the documents we expect to see in any development engagement:
1. A service agreement. It should state explicitly that the intellectual property, the source code and the data of everything developed belong to the client.
2. A Data Processing Agreement (DPA). Under GDPR, the company is the data controller and the provider acts as a processor. The DPA makes that explicit and should also define what happens to the data when the relationship ends.
3. An NDA. Obvious, but we still find it missing more often than you'd think.
4. A transfer clause for conflicts. Any account the provider has opened on the client's behalf must be transferable to the client if it contains their data, code or applications. Including when there's a dispute or litigation.
A contract won't stop a provider from behaving badly. But it turns a hostage situation into a breach of contract with a clear remedy.
Contracts are necessary, but they are not enough. Even with a perfect contract, if the provider holds the keys, enforcing it takes time. Lawyers and courts move in months. Your business needs to keep running independently of the legal disputes and arguments outside of code.
So the practical rule is simple: the client creates the accounts and invites the provider, not the other way around.
For source code, the company should have its own organisation on GitHub (or GitLab, or Bitbucket), create the repositories there and invite the provider's team. When the collaboration ends, you remove their access. Nothing to migrate, nobody to ask.
Infrastructure works the same way. Whether it's AWS, Google Cloud or any other platform, the client opens the account, owns the billing and grants access to the provider.
The same logic applies to everything around the product. A few things worth checking:
- Domain names and DNS
- Apple App Store and Google Play developer accounts
- Third-party services that hold production data (payments, email delivery, analytics, error tracking)
- Credentials and secrets, stored in a password manager the client controls
Sometimes it makes sense for the provider to host and maintain the application. Smaller companies without technical staff often prefer a single supplier that takes care of everything. We do it for some of our clients too.
In that case, sign a proper hosting and maintenance contract, and make sure it covers the exit scenarios explicitly:
- A dispute between both parties
- The provider ceasing its activity
- The client simply wanting to switch providers
In any of them, the provider must transfer the hosting and the data to the client, or to a new provider, whenever it's requested. Within a defined timeframe, and without the service going down during the transition.
Before signing with any development provider, ask yourself one question: if we stopped working together tomorrow, how long would it take us to have full control of our product?
If the answer is anything other than "a couple of hours, revoking a few accesses", fix it now. It's a short conversation at the start of a project, compared with months of disruption at the end of it.
And a provider that's confident in its work has no reason to keep your keys. If someone pushes back on any of this, take it as a warning sign.
If you're already in that situation, talk to our Rescue Squad.

AI is making code cheaper and faster to write, but building software that actually works, scales, and stays secure still takes broad expertise and someone accountable for the outcome. Here's how we see the role of development agencies changing in the AI era, and how MarsBased is adapting to it.
Read full article: The future of development agencies in the AI era
Glocert International has certified our information security management system against ISO/IEC 27001:2022. Here is what the certification covers, what it means for clients, and how to verify it.
Read full article: MarsBased is now ISO/IEC 27001:2022 certified
After months of regional blocking in Spain, we rebuilt our stack to remove Cloudflare dependencies. Using Render and ImageKit.io, we replaced its performance layer without affecting users. So when Cloudflare went down globally on November 18, our site stayed online.
Read full article: How MarsBased website survived Cloudflare's global outage