Most cloud migration decisions are made by what can be signed off this quarter, not by what the workload actually needs. That is how a sensible-looking move becomes a five-year trap.
A lift-and-shift is the cheapest route to start and often the most expensive to live with. Re-platforming changes a little and shifts the economics. Rebuilding is the only honest answer when the application genuinely has to change. Each choice shapes your future cloud bill, your operating model and how easily you could leave.
This post compares all three routes on upfront cost, downtime and long-term flexibility, then gives directors a plain rule for matching each route to the workload in front of them. We will cover why the route you pick this quarter decides your bill for years, what rehosting really moves, where small changes alter the economics, when refactoring is unavoidable, and how the three routes sit side by side. We will also look at sorting workloads before routes, the exit questions that reveal whether a route traps you, where Blowfish fits in the decision, and what to do next. The goal is a cloud migration strategy you can defend in a board pack and live with.
Why the Route You Pick This Quarter Decides Your Bill for Years
Most directors pick the migration route that finishes fastest. Three years later, higher running costs, awkward licensing and limited room to manoeuvre arrive as the invoice. A migration project has a finishing line; a cloud migration strategy does not. It is a commercial decision about operating cost, flexibility and exit options that outlives the project.
Lift-and-shift is widely positioned as the quickest route out of a data centre, with optimisation deferred. Deferred optimisation is a bill you have not seen yet.
So choose by asking which workloads genuinely need to change, not by which route clears the project board this quarter.
This piece defines the three routes, then compares upfront cost, downtime and long-term flexibility. It sorts workloads, tests exit risk and applies one plain rule. If planning is your current stage, see our guide to cloud migration planning for North West businesses.
A reality check: Flexera’s 2024 State of the Cloud findings report 84% of organisations naming cloud spend management as their top challenge (source). [Draft note: verify this figure against the original Flexera report before publishing.]
Lift-and-Shift: What Rehosting Really Moves
Lift-and-shift, or rehosting, moves applications, virtual machines, databases and supporting infrastructure to the cloud with minimal architectural change. The operating model stays as it is. The workload simply changes address.
That is why it is usually the first route considered. It is the closest thing to a straight move, demands the least application work, and gets a business out of a data centre quickly. For small and mid-sized firms, the triggers are familiar: ageing servers, unmaintained software and data centre lease renewals. Directors across the North West will recognise all three.
Be careful with the slogan. The claim that lift-and-shift is always cheapest to start and most expensive to live with is not evidenced in the research. Treat it as a pattern to test against your own workload, not a universal fact.
Where deferred cost tends to surface: over-provisioned instances, unchanged licensing models, unmanaged backups, and no use of the managed services that make cloud economics work.
So treat rehosting as stage one of a migration to the cloud, not a finished migration. Plan the optimisation review before the move completes. For North West businesses that want that review built in from the outset, see How We Deliver Your Cloud Solution.
Re-Platform: Small Changes, Meaningfully Different Economics
Re-platforming goes further than lift-and-shift: you move to the cloud and swap selected components for managed equivalents, such as a self-managed database for a managed database service or a virtual machine for a platform service. The difference in complexity, timeline and cost structure is dramatic, and it shapes long-term operating expense and access to cloud-native capability (Axrail, November 2025).
The payback is practical: less patching, less capacity guessing, fewer late-night database calls, and a smaller estate to secure. The risks are fair to name: a longer project, testing that must be done properly, staff learning new tooling, and provider-specific services that quietly narrow your options.
This is the middle path for most business workloads, which benefit from managed services without justifying a rewrite. Take a line-of-business application with its own database. Re-platforming replaces that database with a managed service and the virtual machines with a platform service. You keep the application logic, but patching, backups and scaling become someone else’s job. When choosing between Azure and AWS for UK SMEs, your cloud migration strategy should match the managed service to the workload, not the brand.
Rebuild: When Refactoring Is the Only Honest Answer
Re-platforming suits workloads that gain from managed services without justifying a rewrite. Rebuild (refactoring or re-architecting) means redesigning an application around cloud-native services rather than reproducing its current shape in a new location.
The justifications are narrow. Rebuild when an application cannot scale past current demand, its data model blocks a committed business change, it is a competitive differentiator, or its current form creates unacceptable exit risk. For context, the GAO found 11 of 69 reviewed legacy systems needed modernisation, with eight on outdated languages and four on unsupported hardware (agencies need to plan for modernising critical legacy systems). That is a genuine rebuild case.
The cost is honest: highest upfront spend, longest timeline, most disruption, and heaviest dependence on the documentation and the people who understand the code. Thin documentation makes estimates guesswork.
Rebuild is the wrong default for back-office systems that already do their job. “We will just rewrite everything” is the most expensive answer to a data-centre exit. Rebuilding only some workloads is normal: a mixed route per workload is a strategy, not a failure of nerve. Rebuild only what must change to meet a business need or genuine exit requirement.
The Three Routes Side by Side: Cost, Downtime and Flexibility
Rebuild is the exception, not the default. Placed side by side, the three routes commit you to very different commercial futures.
| Lift-and-shift | Re-platform | Rebuild | |
|---|---|---|---|
| Upfront cost | Lowest: minimal application change | Moderate: component swaps and testing | Highest: design, build and migration |
| Timeline | Shortest | Middle | Longest |
| Downtime exposure | Driven by data volume, cutover window, dependency mapping, testing and rollback planning. Downtime is a planning outcome, not a property of the route. | Same drivers, more dependencies | Same drivers, greatest complexity |
| Long-term running cost | Often highest without optimisation: over-provisioned instances, unchanged licensing | Lower: managed services cut operational load | Should follow a named business outcome |
| Flexibility | Preserves today’s operating model, limits included | Buys managed services and some native capability | Most room to change |
| Exit options | Ask what leaving costs, what the contract commits you to, and how portable the data is. Proprietary architectures and non-standard formats make exit harder in practice. | Fewer single-provider dependencies | Most portable by design |
A lift-and-shift is cheapest to start and hardest to leave, so the route should follow the workload, not the calendar.
Sort Workloads First, Routes Second
Routes only make sense once you know what you are moving. The 6 Rs is a widely used sorting model: Rehost, Replatform, Refactor, Repurchase, Retire and Retain. Mapping is simple. Rehost is lift-and-shift. Replatform and Refactor sit in rebuild territory, though re-platform leans closest to the middle route. Repurchase, Retire and Retain sit outside the three routes.
A three-question triage, per application
- Must it change to meet a business goal?
- Must it change to stay safe or supported?
- Would changing it unlock a cost or capability we actually want?
Three noes mean keep it as is. From there, making the move in a commercially sensible way begins: Retain a workload on-premises for a genuine reason, or Retire it. Nothing trims a cloud bill like a system you switch off. Repurchase swaps a system for SaaS: procurement, not migration, and often the fastest route to a better position.
Application by application, not house-wide
A single house-wide answer produces either an unbudgeted rebuild programme or a stranded estate of rehosted problems. Name an owner per application, recording route, effort band and review date on one page. That table is the spine of your cloud migration strategy.

The Exit Questions That Decide Whether a Route Traps You
Your workload table gives you a route per application. Before you sign anything, test whether that route can be reversed. Public figures on lock-in are scarce, so treat these as due diligence questions rather than settled facts.
What does data cost to move out? Ask about egress charges, minimum commitments and early termination terms. The CMA’s cloud services market investigation examined egress fees, licensing and discount structures as switching barriers.
How portable is the workload? If it leans on services that exist only in one platform, moving it means rewriting it. Ask now, not in year three.
Who sees the data, and where does it sit? Access, location and retention should align with UK data protection expectations and ICO and NCSC guidance.
What happens at contract end? Get export formats, handover, documentation and a realistic exit timeline in writing. Provider criteria such as those in this IT support buyer’s checklist for Liverpool businesses apply to handover too.
Cloud migration as a service arrangements need particular scrutiny. If a provider runs the environment on your behalf, confirm who controls the data, the configuration and the exit. Ask at route selection, because by then the workload has already taken the shape of the answer.
A Plain Rule for Directors Choosing a Cloud Migration Route
Those answers belong in the decision itself, not in a footnote. So apply one plain rule: keep as-is what does not need to change, re-platform the workloads where managed services remove real operational load, and rebuild only what must change to meet a business or exit requirement.
The rule protects the cheapest, lowest-risk route for stable workloads and reserves the expensive one for cases where it buys something the business can name.
Apply three questions to every workload:
- What business outcome does this route buy?
- What does that outcome cost to run for five years?
- If we wanted to leave in year three, could we?
Cloud migration strategies describe phases; directors need a first move. Start with one or two well-understood workloads, prove the operating model and the support process, then scale. Reliable recovery belongs in that proof, and managed cloud backup for UK businesses explains how to test it properly.
The rule prevents two failures: a blanket lift-and-shift that defers all thinking, and a blanket rebuild that burns budget on systems nobody outside IT will notice changing.
It is a decision filter, not a technical opinion. Agree it with the finance director, not only the IT team.
Where Blowfish Fits in the Decision
The rule above is a decision filter, not a product pitch, and we treat it that way. Blowfish is vendor-neutral on route choice: the recommendation follows the workload, not a service we want to sell. If a rebuild or a full cloud service migration is genuinely the answer, the sensible next step is an assessment conversation, not a quotation.
Two things keep any route honest after the project closes. Cloud Services provides backup and infrastructure, so a migrated estate stays recoverable and supported rather than merely relocated. Managed IT support is the ongoing layer that monitors, patches, secures and answers the phone after cutover, which is the work most migration plans under-cost. Device-level coverage sits in the same conversation, and our Endpoint Protection Service: What North West Businesses Need to Know in 2026 sets out what modern protection looks like for North West SMBs.
Blowfish has delivered managed IT and technology services to North West businesses since 2012, built on 50 plus years of combined experience, so directors get local support rather than a distant ticket queue. For further reading, the NCSC explains how to lift and shift successfully.

What to Do Next
The route you choose sets your future bill and your exit options, so decide by workload need, not by whichever project closes this quarter. Four things to do before you sign anything.
- List and own. Map every workload, name an owner, and mark which genuinely need to change, using the 6 Rs as the sorting lens.
- Cost it honestly. For each workload, compare upfront cost, downtime exposure and five-year running cost, and write down the exit answer before committing.
- Apply the plain rule. Keep what does not need to change, re-platform where managed services remove real load, rebuild only what must change.
- Book the review. Plan the optimisation review before migration day, because an unoptimised estate is an unfinished migration with a monthly invoice.
Naming an owner also settles who maintains the workload afterwards, which is where outsourced and in-house support costs diverge. Our cost comparison for North West SMEs sets out the break-even point.
If the route choice is still open, talk to Blowfish about cloud services, backup and infrastructure, and managed IT support for North West businesses. Then the decision rests on workload evidence rather than vendor pressure.
Conclusion
The route you pick is a financial and operational decision, not a box to tick at quarter end. Sort workloads before routes, because need, not momentum, should drive the choice. Cost each option across upfront spend, downtime exposure and five-year running cost, and write down the exit answer before you commit. Then book the optimisation review early.
Blowfish supports North West businesses with cloud services, backup and infrastructure, and managed IT support, so the decision rests on workload evidence rather than vendor pressure. Our cost comparison for North West SMEs shows where support models diverge. Start by mapping one workload this week, then let the evidence guide the rest. Choose deliberately now, and your cloud estate stays an asset, not a trap.
Frequently Asked Questions
What is the difference between lift-and-shift, re-platforming and rebuilding?
Lift-and-shift (rehosting) moves applications, virtual machines, databases and supporting infrastructure to the cloud with minimal architectural change, so the operating model stays the same. Re-platforming moves to the cloud and swaps selected components for managed equivalents, such as a managed database service. Rebuilding (refactoring or re-architecting) redesigns an application around cloud-native services rather than reproducing its current shape in a new location.
Why does the migration route I choose this quarter matter for years?
A migration project has a finishing line, but a cloud migration strategy does not. The route shapes operating cost, flexibility and exit options long after the project closes. Higher running costs, awkward licensing and limited room to manoeuvre can arrive years later as the invoice, so the route should be chosen by which workloads genuinely need to change rather than by which one clears the project board this quarter.
How should I decide which route fits each workload?
Sort workloads before choosing routes, using the 6 Rs as a sorting model (Rehost, Replatform, Refactor, Repurchase, Retire and Retain). Then apply a three-question triage to each application: Must it change to meet a business goal? Must it change to stay safe or supported? Would changing it unlock a cost or capability we actually want? If all three answers are no, keep it as is. Name an owner per application and record its route, effort band and review date.
What is the plain rule for choosing a cloud migration route?
Keep as-is what does not need to change, re-platform the workloads where managed services remove real operational load, and rebuild only what must change to meet a business or exit requirement. For each workload, ask what business outcome the route buys, what that outcome costs to run for five years, and whether you could leave in year three if you wanted to.
What should I check about exit options before committing to a route?
Test whether the route can be reversed before signing anything. Ask what data costs to move out, including egress charges, minimum commitments and early termination terms. Check how portable the workload is, whether the data is accessible and where it sits, and what happens at contract end, including export formats, handover, documentation and a realistic exit timeline. For cloud migration as a service arrangements, confirm who controls the data, configuration and exit.