Migrating an enterprise application to the cloud costs $50,000 to $150,000 (€44,000 to €130,000, £37,000 to £110,000) for a single application rehosted with a tidy-up, and $300,000 to $1,000,000 (€260,000 to €870,000, £220,000 to £740,000) to replatform a portfolio of five to ten applications. Run costs after migration change by anywhere from minus 20 to plus 30 percent, depending on how much rightsizing is done. Move the least coupled applications first and the core systems of record last, and treat egress fees, licensing and the skills gap as line items from the start.
Prices are in US dollars. Euro and pound figures in parentheses are approximate, rounded conversions. Quotes are issued in the currency of your contract.
Cloud migration is sold as a cost saving and bought as a modernization, and it is sometimes neither. A straight move of the same servers to a cloud provider usually costs more to run, not less, because nothing has been rightsized and the provider is now charging for things the data center absorbed. A migration done well, with applications sequenced, workloads rightsized and the move used as the moment to fix deployment, is the foundation that the rest of a modernization program stands on. This post gives the cost by scope, a sequencing method, an honest view of run costs, the risks that turn up in every post-mortem, and the metrics that tell you whether it worked.
What does cloud migration cost?
| Scope | What is included | Cost | Time |
|---|---|---|---|
| Single application, rehost with tidy-up | Move the application and its database, fix the obvious configuration problems, put a deployment pipeline in place, basic monitoring. | $50,000 to $150,000 (€44,000 to €130,000, £37,000 to £110,000) | 1 to 3 months |
| Single application, replatform | As above plus managed database, containers, secrets management, autoscaling where it helps, and code changes to make those work. | $100,000 to $300,000 (€85,000 to €260,000, £75,000 to £220,000) | 3 to 6 months |
| Portfolio of 5 to 10 applications, replatform | A landing zone (accounts, networking, identity, security baseline), then each application moved in sequence, shared pipelines, central monitoring, cost controls. | $300,000 to $1,000,000 (€260,000 to €870,000, £220,000 to £740,000) | 9 to 18 months |
| Enterprise-wide program | Dozens of applications, a cloud platform team, re-architecture of core systems along the way. | $600,000 to $2,000,000+ (€520,000 to €1,740,000+, £445,000 to £1,480,000+) | 18 to 36 months |
The landing zone is the item most often missed in a first estimate. It is the accounts, networks, identity, logging and security guardrails that every application will share, and it costs $40,000 to $120,000 (€35,000 to €105,000, £30,000 to £90,000) to set up properly before the first application moves. Skipping it means every application builds its own, and the result is unmanageable within a year. The enterprise cost post gives the wider bands these sit inside.
Which applications should move first?
Sequence by coupling and by owner, not by size or by who is shouting loudest.
- First: low coupling, clear owner, low blast radius. Internal tools, reporting, a marketing site, a dev or test environment. These teach the team the platform and prove the pipeline with nothing at stake.
- Second: customer-facing applications with their own data. A portal, an e-commerce front end, a mobile backend. These benefit most from autoscaling and managed services, and the business notices the improvement.
- Third: applications tightly coupled to a core system. Move them once the core has an interface in front of it, otherwise they drag the core with them. The integration post explains that interface.
- Last: the systems of record. The ERP, the ledger, the policy administration system. By now the platform is proven, the team is practiced and the data migration can be rehearsed. Many of these are candidates for a replatform or a SaaS replacement rather than a rehost, and the modernization options should be decided per system before they move.
Two rules cut across the sequence. Move an application together with its database, never one without the other, because a database in the data center talking to an application in the cloud is slow and expensive. And do not move anything that is due to be retired within a year. Retire it instead.
How do run costs actually change?
The honest range is minus 20 to plus 30 percent compared with the data center, and where you land depends on rightsizing.
- A straight rehost runs more expensive. Servers sized for peak load five years ago become instances sized for peak load, billed by the hour, 24 hours a day. Add provider charges for storage, backups, logging and data transfer that the data center never itemized.
- Rightsizing brings it down. Measure actual usage for a month, size instances to it, schedule non-production environments to switch off at night and weekends, and use autoscaling where load varies. This alone is often a 30 to 40 percent reduction from the rehost figure.
- Managed services shift cost from people to provider. A managed database costs more per month than a self-run one and removes a share of a database administrator's time. Whether that is a saving depends on whether the time is redeployed.
- Reserved or committed pricing for stable workloads cuts 30 to 60 percent off list price, once you know what is stable. Commit after six months of real usage, not before.
The saving most companies actually get is not in the cloud bill. It is in the deployment: releases that took a weekend take an hour, environments that took weeks to provision take minutes, and the engineering time that was spent waiting goes into the product. That is worth measuring too.
What are the risks?
| Risk | How it shows up | What to do about it |
|---|---|---|
| Data egress fees | Moving data out of the provider, to another provider or back on premises, is charged per gigabyte. Large archives and backups make this a real number. | Model it before choosing a provider. Keep an exit estimate in the business case. |
| Software licensing | Database and middleware licenses priced per core or per socket behave differently in the cloud. Some vendors charge more, some forbid certain configurations. | Read the license terms for every commercial product before it moves. Budget the difference. |
| Skills gap | The team that ran the data center does not yet run the cloud. Misconfigured permissions and unmonitored costs follow. | Train before the first move, pair with a partner who has done it, and make the landing zone enforce the guardrails. |
| Lock-in | Proprietary managed services are convenient and hard to leave. | Use them deliberately, where the benefit is clear, and keep the application's core logic portable. Containers and standard databases keep options open. |
| Security misconfiguration | A storage bucket left public, an overly broad role, a database open to the internet. | Guardrails in the landing zone, automated checks on every deployment, and a review before each application goes live. |
| Uncontrolled cost | Nobody owns the bill. It grows every month and nobody can say why. | Tag everything by application and owner from the first day. Budgets with alerts. A monthly review with the owners in the room. |
Compliance deserves its own line. Data residency, encryption, audit logging and access control all have to be designed into the landing zone, not retrofitted. Building software for regulated industries covers what changes when the regulator is a stakeholder.
What should you measure after migration?
- Cost per application per month, tagged, against the pre-migration baseline and the business case.
- Deployment frequency and lead time, from commit to production. This is where the value usually appears.
- Availability and incident count, compared with the twelve months before.
- Time to provision an environment, from request to usable.
- Rightsizing progress: the share of instances that have been resized or scheduled since the move.
- Security findings from automated checks, trending down.
If cost is the only metric, the migration will be judged a failure in month three, before rightsizing and before the deployment gains are visible. Set the full list of metrics before the first application moves and report them monthly.
Before the first application moves
- Build the landing zone: accounts, network, identity, logging, guardrails, tagging.
- Sequence the portfolio by coupling and owner. Retire what you can instead of moving it.
- Read the license terms for every commercial product that moves.
- Set the baseline for cost, deployment lead time and availability.
- Decide who owns the bill and who reviews it monthly.
How 7L approaches cloud migration
7L International is a custom software development and digital transformation firm with business hubs in New York, London and Athens, founded during the Greek financial crisis, and we have supported more than one hundred organizations, including through migrations where the cloud move was the first slice of a wider modernization. We build the landing zone first, move the low-risk applications to prove it, and use each move to put a real deployment pipeline in place, because that is where the return is. Core systems move last, usually with an interface in front of them and a decision already taken about whether they are being rehosted, replatformed or replaced. The enterprise software development guide covers how that fits a full program. If you are planning a move and the business case is only about the bill, tell us what you run today and we will help you build the case that includes the rest.
Frequently asked questions
How much does it cost to migrate an application to the cloud?
A single application rehosted with a tidy-up, including a deployment pipeline and basic monitoring, costs $50,000 to $150,000 (€44,000 to €130,000, £37,000 to £110,000) and takes one to three months. Replatforming it onto managed services costs $100,000 to $300,000 (€85,000 to €260,000, £75,000 to £220,000). A portfolio of five to ten applications, with a shared landing zone, costs $300,000 to $1,000,000 (€260,000 to €870,000, £220,000 to £740,000) over nine to eighteen months. The landing zone alone is $40,000 to $120,000 (€35,000 to €105,000, £30,000 to £90,000) and is the item most often left out of first estimates.
Does moving to the cloud reduce costs?
Not by itself. A straight rehost usually costs more to run, because instances are sized for old peaks and the provider itemizes storage, backups, logging and data transfer. Rightsizing, scheduling non-production environments off out of hours and committed pricing for stable workloads bring it down, and the realistic range after that work is minus 20 to plus 30 percent against the data center. The larger saving is usually in deployment speed and engineering time.
Which applications should be migrated first?
Applications with low coupling, a clear owner and a small blast radius: internal tools, reporting, test environments. Then customer-facing applications with their own data, which benefit most from autoscaling. Applications tightly coupled to a core system move once the core has an interface in front of it. Systems of record move last, after the platform and the team are proven. Anything due for retirement within a year should be retired, not moved.
What is a cloud landing zone?
The shared foundation every application sits on: account structure, networking, identity and access, logging, security guardrails, cost tagging and budgets. It is built once, before the first application moves, and it is what makes a portfolio of applications manageable rather than a collection of individually configured accounts. It costs $40,000 to $120,000 (€35,000 to €105,000, £30,000 to £90,000) to set up properly for a mid-sized company.
What are the biggest risks in a cloud migration?
Uncontrolled cost from untagged resources nobody owns, security misconfiguration such as public storage or overly broad roles, software licenses that cost more or are not permitted in the cloud, data egress fees that make leaving expensive, and a skills gap in the team that ran the data center. Each is manageable if it is a line in the plan from the start, and each is a post-mortem finding when it is not.