Vendor lock-in rarely feels like a problem on the day you sign. The cloud credits are generous, the SaaS tool does what you need, the agency that built your app is responsive. The bill arrives later: when prices rise, when the vendor is acquired, when the agency stops answering, or when you simply want to move and discover that moving will take six months and a rebuild.
This article gives you a practical way to measure that hidden bill before you sign, and the contract clauses and architecture choices that keep it small.
What does vendor lock-in in cloud computing actually mean?
Lock-in is not the same as dependence. Every business depends on vendors. Lock-in is when the cost of leaving becomes so high that you can no longer negotiate, even if the service gets worse or more expensive. It comes in four forms:
- Data lock-in: your records sit in a proprietary format or service, and getting them out is slow, partial or expensive.
- Technical lock-in: your application is built on services that only exist at one provider, such as a proprietary database, queue or serverless runtime, so moving means rewriting.
- Commercial lock-in: multi-year commitments, prepaid credits, or bundled discounts that penalise leaving.
- Knowledge lock-in: only the agency or one freelancer understands how the system is deployed, where the credentials are and how it fits together. This is the most common form in Indian SMEs, and the least discussed.
Is lock-in getting better or worse?
There has been some movement in the customer's favour. In January 2024 Google announced it would waive data transfer fees for customers leaving its cloud, and on 5 March 2024 AWS followed, offering credits for data transfer out to the internet for customers moving their data off AWS, after contacting AWS to request them. Reporting at the time linked both moves to the European Data Act, which aims to make switching cloud providers easier.
Two cautions. First, those waivers cover the transfer fee for a full exit; they do nothing about the engineering cost of rewriting an application tied to proprietary services, which is usually the larger cost. Second, terms change, so check the provider's current conditions before you rely on them.
The exit-cost test: run it before you sign
For any significant vendor, cloud, SaaS or development agency, answer these ten questions and write the answers down. If you cannot answer one, that is your risk.
- Can we export all our data, including attachments, history and audit logs, in an open format (CSV, JSON, SQL dump), on demand, without raising a ticket?
- How long would a full export take, and has anyone actually tried it?
- Which features of this product would we have to rebuild elsewhere? List them.
- Who owns the source code, designs and deployment scripts? Is that written in the contract?
- Are the domain name, DNS, app store accounts and cloud accounts registered in our company's name, with our email as owner?
- Do we hold the admin credentials ourselves, or only the vendor does?
- What notice and assistance does the contract promise when we leave?
- What are the price-increase terms after the first year?
- If the vendor shuts down tomorrow, what do we have in our hands?
- Roughly how many person-weeks would it take another team to take over?
Turn the answers into a rough rupee figure: engineering weeks to rebuild or migrate, plus downtime cost, plus any contractual penalty. That number is your exit cost. Compare it with the annual contract value. When exit cost is several times the yearly spend, you are locked in, whatever the brochure says.
Which contract clauses protect you?
| Clause | What it should say | Why it matters |
|---|---|---|
| Data ownership | All data and derived records belong to the customer | Prevents disputes over what you may take |
| Data export | Full export in a documented open format, within a fixed number of days, at no extra fee | Turns "we will try" into an obligation |
| Exit assistance | A defined transition period with continued service and reasonable support | Stops a hard cut-off mid-migration |
| IP and source code | Code, designs and scripts assigned to the customer on payment | Essential for agency-built software |
| Accounts and credentials | All cloud, domain, store and third-party accounts in the customer's name | Removes knowledge lock-in |
| Price protection | Cap on annual increases, notice period for changes | Limits commercial lock-in |
| Data deletion | Certified deletion after exit | Supports the storage-limitation principle under India's DPDP Act |
If a vendor refuses a reasonable export clause, treat that as information about how leaving will go.
Which architecture choices keep you portable?
You do not need to avoid managed cloud services; they save real effort. The goal is to use them where switching is cheap and be careful where it is not.
Usually safe to use
- Managed versions of open-source engines: PostgreSQL, MySQL, Redis, Kubernetes, object storage with an S3-compatible API. Every major provider and many Indian hosts offer equivalents.
- Containers. An application packaged as Docker images runs on any cloud, a VPS or your own server.
- Infrastructure as code (Terraform, OpenTofu, Ansible), kept in your own Git repository.
Use with a plan
- Proprietary databases, serverless functions with provider-specific triggers, and provider-only AI or analytics services. Use them when they genuinely save time, but keep them behind a small internal interface in your code so that only one module changes if you move.
- Identity and login. Use standards such as OpenID Connect and SAML so users can move with you.
Avoid
- Business logic written inside a no-code tool or a vendor's scripting language with no export path.
- A single cloud account owned by the agency's email address.
The US cybersecurity agency CISA, in its ransomware guidance, even suggests considering a multi-cloud approach for cloud-to-cloud backups partly to reduce vendor lock-in. A second copy of your data with a different provider is cheap insurance on both counts.
SaaS lock-in: the tools you rent
CRM, HRMS, accounting and helpdesk tools are where data quietly accumulates. Three habits help:
- Schedule a monthly automated export of key objects (customers, deals, invoices, tickets) into storage you control.
- Avoid building critical approval flows only inside a SaaS tool's automation builder; document them outside too.
- Before renewing, ask the exit-cost questions again. Renewal is the moment you have the most bargaining power.
Agency lock-in: the one founders feel most
A common Indian story: an agency builds the app, hosts it on its own server, keeps the code in its own repository, and the relationship sours. The business discovers it owns a URL and nothing else. Prevent it by insisting from the first invoice that code lives in a repository your company owns, hosting is in your account, and a short handover document lists every service, credential and deployment step. If you are already in that position, recover the accounts first; a maintenance plan with a new team can then keep the system running while you take back control.
A simple portability checklist
- Company-owned cloud, domain, DNS and app store accounts, with two internal admins.
- Code in a company-owned Git repository, with a readme that explains deployment.
- Containers and infrastructure as code where practical.
- A tested monthly data export from every SaaS tool that holds customer data.
- One provider-specific dependency list, reviewed every year.
How Cehpoint can help
Cehpoint builds on the principle that you own the result: code, infrastructure and IP. We can audit your current setup for lock-in, move accounts and code into your name, and containerise applications so they can run on any cloud or on your own servers. Our cloud and DevOps service covers deployment and hosting, and if rising bills are the trigger, a cloud cost review is a practical place to start.