How do I avoid cloud lock-in and get a proper handover?

How small D2C brands can avoid lock-in to a cloud vendor, agency or developer, and what a complete technical handover should include.

Anadoris Digital · · 4 min read

Metal chain
Photo: JJ Ying / Unsplash (CC0)

Lock-in worries small brands in two ways. The first is vendor lock-in: building so deeply on one cloud's special services that moving becomes impractical. The second, and usually more painful, is people lock-in: depending on one developer or agency who holds the passwords, knows how things work and has never written any of it down.

Some dependence is unavoidable and even sensible. Using managed services saves real work. The aim is not to avoid every provider-specific feature, but to make sure that choosing to move, or a person leaving, is a manageable project rather than a crisis.

Ownership is the foundation

  • Cloud accounts, domains and DNS registered to the company.
  • Code repositories in a company organisation, not personal accounts.
  • Third-party services, such as email, payment, courier and monitoring tools, in company accounts.
  • Named access for each developer or agency, removable at any time.
  • Contracts stating that the brand owns code, data and documentation created for it.

Without ownership, nothing else matters. With it, even a difficult parting is mostly a matter of changing passwords and reading documents.

Reducing technical lock-in sensibly

Using AWS services such as Lambda, DynamoDB or SQS is not a mistake; they often save time and money. The trick is to use them knowingly and keep the most important parts portable.

  • Prefer standard databases such as PostgreSQL or MySQL for core business data.
  • Package apps in containers or common frameworks that run on many platforms.
  • Define infrastructure in code with tools such as Terraform, CDK or SAM, so it is documented by default.
  • Keep business logic separate from provider-specific glue, so the glue can be replaced.
  • Export data regularly in open formats such as CSV or standard database dumps.

Documentation that stays useful

Documentation does not need to be long. It needs to be current and written for the next person. A short set of living documents, kept in the code repository or a shared wiki the brand owns, covers most needs. Ask for documentation as part of the work, not as a separate project at the end, because end-of-project documentation is the first thing cut when time runs short.

  • An architecture diagram showing systems, data flows and where each runs.
  • A system list with owners, accounts, regions and key settings.
  • Runbooks for deploying, rolling back, restoring backups and handling common alerts.
  • A secrets inventory listing each key, its purpose and where it is stored, without the values.
  • A record of important decisions and why they were made.

Handover checklist

  • All accounts confirmed in the company's name, with the brand holding root or owner access.
  • Outgoing team's access removed, and secrets they could see rotated.
  • Architecture diagram, system list and runbooks reviewed and up to date.
  • Infrastructure code and app code in company repositories, building successfully.
  • Backups confirmed and a restore tested by the incoming team.
  • Monitoring and alerts pointed to the new owners.
  • Open issues, known risks and upcoming renewals listed.
  • A recorded walkthrough session for future reference.

Test the handover before you rely on it

The real test of a handover is whether the new team can work without calling the old one. Ask the incoming team to deploy a small change, restore a backup into staging and answer a few questions using only the documents. Where they get stuck, the outgoing team fills the gap and updates the documents. Agree a short overlap period in the contract for exactly this, and schedule the test early in that period.

What to put in contracts

Contracts are the easiest place to prevent lock-in, and the hardest to fix afterwards. Whether you hire a freelancer, an agency or a managed provider, the agreement should state that the brand owns the code, data, documentation and accounts created for it, that documentation is maintained as part of the work, and that a defined handover, including an overlap period, happens when the engagement ends.

It should also cover access: who has it, how it is granted and how it is removed. Ask for any third-party licences or tools to be in the brand's name where possible. None of this signals distrust. Good partners welcome clear terms, because they protect both sides and make the eventual handover routine rather than tense.

Where to start

Check that every account and repository is in the company's name today. Ask your current developer or agency for an architecture diagram, a system list and a secrets inventory, and review them together. Add a handover clause to future contracts covering documentation, access and an overlap period. Doing this while relationships are good is far easier than during a dispute. Revisit the documents every few months, and especially before a major sale, so they never fall far behind the systems they describe.

Anadoris Digital works this way by default, with the brand owning every account and documentation maintained as part of the work.

Related service: AWS & Cloud Management · More on this topic: AWS & Cloud Management articles

Common questions

Is using AWS-specific services a mistake?

No. Managed services such as Lambda, SQS or DynamoDB often save time and money. Use them knowingly, keep core data in standard formats, define infrastructure in code and keep business logic separate, so moving later is a project rather than a rewrite.

What should be in a technical handover document?

An architecture diagram, a list of systems and accounts, runbooks for deploys, rollbacks and restores, a secrets inventory without values, open issues and known risks, and upcoming renewals. Pair it with a walkthrough session and a short overlap period.

How do I get my code back from a developer?

Ask for the repository to be transferred to your company organisation, or for full access to an exported copy including history. If the contract states the brand owns work created for it, refer to that. Going forward, keep code in company-owned repositories from day one.

How long should a handover overlap be?

It depends on how complex your systems are. The point is to allow the incoming team to deploy, restore and handle a routine issue with the outgoing team available. Agree the overlap in the contract rather than leaving it to goodwill at the end.

Your next chapter

Let’s build what comes next.

A new website. A better system. A simpler way to work. Tell us where you want to begin.

Discuss your project