Application Migration Roadmap: 6 Phases and Owners
9 min read ยท 2026-11-20
An application migration roadmap is the plan that sets which applications move, in what order, with which strategy and by what date. It runs in six phases: inventory, assessment and dependency mapping, a strategy decision for each app, a pilot wave, the main migration waves, then decommissioning and optimization.
A checklist lists tasks. A roadmap does more: it groups applications into waves, gives each wave an owner, and sets the gates a wave must pass before the next one starts. The template below covers about nine months for a portfolio of a few dozen applications. Add or remove waves to fit your size.
The roadmap at a glance
Goal: Move every in-scope application to its target platform in planned waves, with no unplanned downtime, and shut down the old environment. Duration: About 9 months
Mobilize and inventory (Weeks 1-4)
Know exactly what you are moving and who decides.
- Name a migration lead, an executive sponsor and one business owner for each application.
- Build the inventory: application, owner, users, data size, hosting, licenses, support contracts and end dates.
- Pull usage data so you can spot apps nobody uses anymore.
- Agree on scope, target platform and the business reason for the move.
Milestone: Inventory signed off by the sponsor, every application has a named owner.
Assess and map dependencies (Weeks 3-10)
Understand how hard each move is and what breaks if you move one app alone.
- Score each app on complexity, criticality, compliance needs, availability requirements and how easily it can be tested.
- Map dependencies: shared databases, APIs, batch jobs, file shares, single sign-on, hard-coded IP addresses.
- Group tightly linked apps into "move groups" that have to migrate together.
- List compliance constraints such as HIPAA, GDPR or PCI DSS for each app.
Milestone: Dependency map and assessment scores reviewed with each application owner.
Decide strategy and plan waves (Weeks 9-12)
Pick one strategy for each app and put the move groups in order.
- Assign one of the 7 Rs to each app, using the decision rules below.
- Build waves: easy, low-risk apps first, critical apps once the method is proven.
- Set cutover windows with the business, outside peak periods and freezes.
- Write the entry and exit criteria for every wave.
Milestone: Wave plan and budget approved by the sponsor.
Pilot wave (Months 4-5)
Prove the method on two to four low-risk applications.
- Build the target landing zone: network, identity, security, monitoring and backup.
- Migrate the pilot apps, run functional and performance tests, then cut over.
- Run the rollback once, for real, on one app.
- Write down what took longer than planned and update the runbook.
Milestone: Pilot apps live on the target for two weeks with no rollback needed.
Migration waves (Months 5-8)
Move the rest of the portfolio, one wave at a time.
- Freeze changes on each app a set number of days before its cutover.
- Sync data, test, cut over, then run old and new in parallel for an agreed period.
- Hold a go or no-go meeting before every cutover.
- Review each wave before the next one starts and change the plan where needed.
Milestone: All waves cut over and business owners have signed acceptance.
Decommission and optimize (Months 8-9)
Turn off the old environment and capture the savings.
- Before you shut down an app marked for retirement, run a controlled stop: block its traffic or pause it for a set period and watch what breaks.
- Archive data as your retention rules require, then shut down the source servers.
- End the old licenses, hosting and support contracts on their notice dates.
- Right-size the new resources based on real usage.
- Hand over to operations with updated documentation and runbooks.
Milestone: Source environment switched off and old contracts terminated.
Why application migrations stall
Most migrations don't fail on the technology. They stall because of a few planning gaps that show up halfway through. Here are the four most common, and the phase of the roadmap that fixes each one.
- Dependencies found too late. A wave is ready to cut over when someone notices that another app reads the same database or calls an API on the server. The wave stops. Fix it in phase 2, Assess and map dependencies, by building move groups before you plan any wave.
- A rollback that was never tested. The plan says "roll back if needed", but nobody knows how long it takes or who decides. Fix it in phase 4, Pilot wave, by running a real rollback on one app.
- Unclear owners. Nobody can confirm the cutover window or sign acceptance, so dates slip week after week. Fix it in phase 1, Mobilize and inventory, by naming a business owner for every application.
- Moving apps nobody needs. Time and budget go into apps with almost no users. Fix it in phase 3, Decide strategy and plan waves, by retiring them before the waves start.
What an application migration roadmap is, and what it is not
The roadmap is a sequencing document. It answers three questions for every application: what happens to it, when, and who is accountable. The technical runbook for each cutover sits under it. It is not the roadmap.
This difference matters when you present the plan. Leadership needs the waves, dates, risks and decisions. Engineers need the runbooks. Put both in one document and neither group reads it. Keep the roadmap to one page and link the detailed work from each wave.
How to choose a strategy for each application
The Guide for AWS large migrations, part of AWS Prescriptive Guidance, describes seven migration strategies, often called the 7 Rs. Give each app exactly one, using these rules.
Refactoring is the most complex strategy and is hard to manage across many applications. Rehost, relocate or replatform first, then modernize after the migration. In practice: migrate first, modernize after. Put refactoring in a later phase of the roadmap so it does not block the waves.
- Retire when the app has no business value left. Usage data helps you spot candidates: low usage over a set period, or no inbound connection over a set period. Set your own threshold and window from your discovery tool data, not from memory.
- Retain when the app can't move yet: a contract still running, a pending replacement or a hard compliance limit.
- Rehost (lift and shift) when the app works fine and the goal is speed.
- Relocate when you move whole virtualized workloads to the cloud version of the same platform without changing them.
- Replatform when a small change, such as a managed database, brings a clear gain without a rewrite.
- Repurchase when a SaaS product does the job and the custom app costs more to keep.
- Refactor or re-architect only when the business needs features the current design cannot deliver.
Who owns each phase
Use four roles at minimum. The sponsor approves the scope, budget and go or no-go calls on critical apps. The migration lead owns the plan, the waves and the risk log. The technical lead owns the landing zone, tooling and runbooks. Each business owner confirms the cutover window and signs acceptance.
Write the owner on every phase and every wave of the roadmap itself, not in a separate RACI file nobody opens. When a date slips, the reader should see at once who to ask.
Cutover, rollback and parallel run
Every wave needs a defined point of no return. Before it, rolling back means pointing users at the old system again. After it, new data exists only on the target and rolling back means losing or reconciling that data. Mark this point on the timeline and hold the go or no-go meeting just before it.
For each cutover, write down:
- The rollback trigger: which failed test or error level sends you back.
- Who can call the rollback, and how long it takes.
- How long old and new run in parallel, and when the source becomes read-only.
- How data written during the parallel run gets reconciled.
How much does an application migration roadmap cost?
There is no honest single number. The cost depends on how many apps you have, which strategy each one gets and how long you run two environments. Build the budget line by line from the roadmap instead of starting from a total.
The main lines are: target infrastructure, migration tooling, licenses (new ones plus old ones still running during the overlap), people (internal time and any partner), testing, training, and double running costs during parallel runs. Retiring apps and ending old contracts brings the cost down. Refactoring, long parallel runs and late discovery of dependencies push it up. Show the budget per wave, so leadership sees spending and savings follow the timeline.
Example: applying the six phases to an email migration
An Exchange or email migration is a good way to see the six phases at a smaller scale. Each step below maps to one phase of the roadmap.
- Inventory: list mailboxes, shared mailboxes, distribution lists and the apps that send mail through the server.
- Assess: check that the source version supports the migration method you picked, such as cutover or hybrid.
- Decide: remove inactive mailboxes and cut oversized ones down before moving them.
- Pilot: prepare DNS records, certificates and identity sync, then move a pilot group.
- Waves: migrate the rest in batches, VIP users last, then switch mail flow.
- Decommission: keep the old server for a set period, then shut it down.
How to present the roadmap to leadership
Use two views. The one-page view shows the six phases on a timeline, the waves as bars, owners on each bar, and milestones as diamonds. Add the top three risks and the decisions you need. The detailed view lists every app with its strategy, wave and cutover date.
Review the roadmap every quarter, or after each wave on shorter programs. Keep one reusable cloud migration template so the next wave, or the next project, starts from a proven structure instead of a blank page.
Common mistakes to avoid
- Ordering waves before mapping dependencies stalls cutovers, so map shared databases and APIs first, then build waves from move groups.
- Refactoring during the move slows every wave, so rehost or replatform first and put modernization in its own later phase.
- Skipping the rollback test leaves you guessing on cutover day, so run a real rollback during the pilot wave.
- Migrating apps nobody uses wastes time and budget, so check usage data and retire them before the waves start.
- Retiring an app without a controlled stop hides dependencies, so block its traffic for a while first and let them show up before the server is gone.
- Forgetting to end old contracts delays the savings, so put notice dates on the roadmap.
- Treating it as an IT-only project lets dates slip, so give each app a business owner who signs acceptance.
Frequently asked questions
What are the phases of an application migration roadmap?
Six phases: mobilize and inventory, assess and map dependencies, decide strategy and plan waves, pilot wave, migration waves, then decommission and optimize. Assessment and planning overlap a little so you can start on the waves sooner.
How long does an application migration take?
It depends on the number of apps, their dependencies and the strategies you choose. The template above covers about nine months for a few dozen applications. Larger portfolios add waves, and refactoring adds much more time.
What is the difference between rehost and replatform?
Rehost moves the application as it is to new infrastructure. Replatform makes small changes along the way, such as switching to a managed database, without rewriting the application.
Which applications should you migrate first?
Start with low-risk, low-dependency applications that are easy to test. They prove the method and the tooling. Move critical, tightly linked systems once your runbook has worked in at least one wave.