Cloud modernisation ROI: how to estimate the cost and the benefit
A step-by-step way to estimate the return on a cloud modernisation, with a worked example in round numbers and the costs that are easiest to miss.
By Sandeep ThesiyaChief Technology Officer (CTO), Coretus Technologies
Updated
Short answer
To estimate cloud modernisation ROI, compare the total cost of staying as you are with the total cost of migrating and running in the cloud over the same period. Count one-off migration and dual-running costs, ongoing cloud and staff costs, and only savings you can prove. Show faster releases and lower risk as separate, softer benefits.
Cloud modernisation ROI is the net benefit of moving and improving your systems, divided by what the move costs, over a fixed period, typically three to five years. You build it by comparing two scenarios: carrying on as you are, and migrating. Count every one-off and running cost on both sides, and keep hard savings separate from softer benefits such as faster releases.
The formula
Keep the calculation simple enough for a finance team to check line by line:
- Savings = what you would spend staying as you are, minus what you will spend running in the cloud, over the same period. Include costs you avoid, such as a hardware refresh.
- Net benefit = savings minus the one-off migration costs.
- ROI = net benefit divided by the one-off migration costs.
- Payback point = when cumulative savings first cover the one-off costs.
If your finance team prefers net present value, discount each year's figures at their usual rate. The inputs stay the same.
Step 1: Baseline what you spend today
This is often the weakest part of a business case, because the costs sit in different budgets. Collect:
- Hardware support, depreciation and the date of the next refresh.
- Data centre or colocation fees, including power and cooling.
- Software licences tied to servers or processor cores, and support contracts.
- Staff time spent on patching, backups, capacity planning and hardware faults. Ask for hours, then convert them to cost.
- Downtime and incidents: how often, how long, and what an hour of downtime costs in lost sales or idle staff.
- Delay costs: releases held back by slow environments or risky deployments.
Step 2: Choose a migration approach for each application
The approach changes both the one-off cost and the savings. AWS groups the options into seven strategies, often called the 7 Rs: retire, retain, rehost, relocate, repurchase, replatform, and refactor or re-architect. Rehosting, or lift and shift, is usually the quickest, but it brings no cloud optimisations by itself. Refactoring is the most complex and costly, and is chosen when the business needs to scale, release faster or cut running costs.
Decide application by application. A case that assumes everything gets refactored will be too expensive. One that assumes everything is lifted and shifted often shows little saving, because oversized servers move to the cloud unchanged. Some applications should simply be retired.
Step 3: Price the one-off costs
- Assessment and planning.
- Migration engineering: refactoring, replatforming, testing and cut-over.
- Data migration and validation.
- Dual running, which means paying for the old and new environments at the same time while you migrate.
- Training for the people who will run the new platform.
- Contingency for what you find once the work starts.
Step 4: Estimate the running costs in the cloud
- Compute and storage, sized from real usage rather than from today's server specifications.
- Data transfer, especially traffic leaving the cloud provider, which is easy to miss.
- Managed services such as databases, queues and load balancers.
- Monitoring, security tools, backups and support plans.
- Licences that remain, and any that change price in the cloud.
- Staff time to run the platform and keep the bill under control. Our cloud cost optimisation work focuses on this part.
Step 5: Value the benefits honestly
Split the benefits into two groups and show them separately.
Hard savings are costs that stop: hardware refreshes you no longer buy, colocation contracts you end, licences you drop. Finance can verify these.
Soft benefits are real but harder to prove: fewer incidents, faster releases, staff time moved to product work, and the ability to handle peaks. Staff time only becomes a cash saving if spending actually falls. Otherwise it is freed capacity, which is still worth stating as long as it is labelled.
Put a money value on a soft benefit only when you can tie it to something measurable, such as the revenue from launching a feature earlier, and mark it clearly as an estimate.
A worked example in round numbers
This example is hypothetical. The figures are round numbers chosen to show the method. They are not client data, benchmarks or a quote.
A company runs its main applications in a colocation facility. It compares three years of staying put with three years of migrating.
Scenario A: stay as you are
- Running costs of $400,000 a year for hosting, hardware support, licences and staff time on infrastructure.
- A hardware refresh of $250,000 due in year two.
- Three-year total: $1,450,000.
Scenario B: migrate
- One-off costs of $420,000: migration engineering $300,000, dual running $100,000 and training $20,000.
- Running costs of $320,000 a year: cloud bill $180,000, remaining licences $50,000, monitoring and tools $20,000, and staff time $70,000.
- Three-year running total: $960,000.
The result
- Savings: $1,450,000 minus $960,000 = $490,000.
- Net benefit: $490,000 minus $420,000 = $70,000.
- ROI over three years: $70,000 divided by $420,000, or about 17%.
- Payback point: early in year three, when cumulative savings pass $420,000.
That is a modest but positive case, built on hard savings alone. Now test it. If the cloud bill comes in 20% higher than planned, at $216,000 a year, running costs rise to $356,000 a year. Savings fall to $382,000, the net benefit becomes minus $38,000, and the project no longer pays for itself within three years on hard savings alone.
This is why sizing cloud resources from real usage, and planning cost controls from the first day, matters as much as the migration itself. It is also why soft benefits belong in the case. If the move lets the team release monthly instead of quarterly, and an earlier launch is worth more than $38,000, the decision changes. Show that as a separate line, with its assumptions written next to it.
Costs that are easy to forget
- Dual running that lasts longer than planned.
- Data transfer charges between regions and out to the internet.
- Oversized servers moved unchanged, so the cloud bill mirrors old capacity.
- Licences that cost more in the cloud, or aren't allowed there at all.
- Monitoring, logging and security tools that came bundled with the old setup.
- Staff time to learn and run the new platform in the first year.
What a phased modernisation looked like
For a global core banking client, we moved the system from an ageing mainframe to AWS one function at a time, keeping both ledgers in sync throughout. The board had ruled out a single switch-over as too risky, so the move was planned in phases from the start. The client saw a 60% operating cost reduction, 5x release speed and zero service downtime.
Where to start
List your applications and, for each one, the migration approach that fits and what it costs to run today. That inventory is most of the work. Then fill in the two scenarios above with your own figures, and test them against a cloud bill 20% higher than planned.
If you want a second opinion on the numbers, our legacy-to-cloud migration team can review them with you. If the cloud move is part of a wider change programme, read our step-by-step digital transformation plan.
Sources
- About the migration strategies, AWS Prescriptive Guidance

About the author
Sandeep Thesiya, Chief Technology Officer (CTO)
Chief Technology Officer at Coretus Technologies. Sets the technology vision and engineering direction: scalable platforms, production AI systems and cloud infrastructure.