"How long will this take?" is usually the first question a business owner asks about a cloud migration, and "it depends" is the least satisfying answer anyone can give. It is also the honest one. The good news is that "it depends" breaks down into a handful of specific things you can look at today, before anyone signs a statement of work.
This post walks through the four phases every migration goes through, what tends to stretch each one, and how to build a schedule you can actually hold to.

AWS, Azure, and Google Cloud can all spin up servers in minutes. The cloud side is rarely the slow part. What takes time is understanding what you already have: how many applications, how they talk to each other, how much data sits behind them, and who has to sign off before anything changes.
That's why two companies of the same size can have wildly different timelines. A business running a few well-documented servers can move quickly. A business with fifteen years of accumulated systems, a few custom integrations, and an app that only one retired contractor understood will spend most of its time on discovery, not on the move itself.
Assessment answers three questions: what do we run, what depends on what, and who owns it? You list every server, database, file share, scheduled job, and SaaS integration, then map the connections between them.
This phase almost always turns up surprises. A reporting server that a single department still relies on. A nightly script that copies files to a machine under someone's desk. A license tied to a specific piece of hardware. Every surprise found here is cheap. Every surprise found during cutover is expensive.
Assessment goes fastest when someone can pull a clean inventory from existing tools and the people who know the systems are available to answer questions. It drags when documentation is thin and those people are busy with their day jobs.
Not every workload should move the same way. For each one, you pick a strategy:
The plan also covers the landing zone (accounts, networking, identity, security baselines), the order workloads will move in, and how you'll roll back if something goes wrong. Skipping the landing zone to "save time" is one of the most reliable ways to lose time later.
The temptation is to move everything in a single heroic weekend. Resist it. Moving in waves — small groups of related workloads at a time — lets you learn from each round, fix your runbook, and keep the blast radius small if something misbehaves.
A sensible first wave is something low-risk but real: an internal tool or a test environment. It proves the network, identity, backups, and monitoring work before anything customer-facing moves. Later waves go faster because the team has already hit and fixed the common problems.
Each wave follows the same pattern: replicate data, test in the new environment, schedule a cutover window, switch over, and keep the old system available until you're confident you won't need it.
Plenty of migration plans end at cutover. Real migrations don't. The first few weeks in the cloud are when you right-size instances that were provisioned generously "just in case," tune backups and alerting, tighten permissions, and check the first full bill against what you expected.
If this phase isn't in the timeline, it doesn't happen — and the result is a cloud environment that costs more than it should and that nobody fully trusts. Budget time for it from the start.
When a migration runs long, it's rarely because the technology failed. It's usually one of these:
These are ballpark ranges from the kinds of projects small and mid-sized businesses typically take on — not promises, and not a substitute for an assessment. Use them to sanity-check a plan, not to set a launch date.
| Scenario | What it usually looks like | Rough timeline |
|---|---|---|
| A handful of servers | A few well-understood workloads, mostly rehosted, modest data, one or two owners | A few weeks |
| A typical SMB environment | A mix of line-of-business apps, a database or two, file shares, some integrations | A few months |
| Apps that need rework | Key systems being replatformed or refactored, with testing and user acceptance | Several months, sometimes longer |
| A full data center exit | Many workloads, large data volumes, compliance requirements, multiple waves | Many months to a year or more |
Notice the pattern: the timeline grows with complexity and risk, not with headcount or company revenue.
Before you commit to a date, make sure you can check most of these off:
Plenty of teams migrate a few servers on their own and do it well. A partner earns their keep when the stakes or the unknowns go up: when you don't have a full picture of what you run, when downtime directly costs you customers, when compliance is involved, or when your team simply can't step away from daily operations long enough to run a project.
A good partner shortens the timeline mostly by shortening the surprises. They've seen the hidden dependency and the licensing snag before, so they look for them early instead of finding them at 2 a.m. on cutover night.
A cloud migration takes as long as it takes to understand what you have, decide what to do with each piece, and move it carefully. Do the first two well and the third goes faster than you'd expect.
Cloudology plans and runs migrations to AWS, Azure, and Google Cloud through our cloud migration practice, starting with an assessment that gives you a timeline grounded in your actual environment. If you're trying to figure out how long your move will take, let's talk through what you're running.