Top Challenges of Cloud Migration and The Best Solutions
September 22, 2026
September 22, 2026
The server move went fine on Friday night. Then Monday happened.
QuickBooks can’t find the company file. The label printer in shipping has vanished. Three people can’t open the CRM, because nobody knew its license server was running on an old PC under the reception desk, and that PC is now unplugged in a storage room.
None of this is exotic. Most cloud migration challenges aren’t technical mysteries. They’re things nobody wrote down, costs nobody priced in, and people nobody asked. They also hit small businesses harder, because there’s usually one IT person (or one outside consultant) holding the whole project together.
The data backs this up. A McKinsey survey of nearly 450 CIOs and IT leaders found the average company spent 14% more on migration than planned each year, and 38% saw delays of more than one quarter (McKinsey, 2021). That survey is a few years old, but the pattern hasn’t gone anywhere. In Flexera’s 2026 State of the Cloud Report, 85% of organizations still name managing cloud spend as a top challenge (Flexera, 2026).
Below are the eight problems that show up most often when small and midsize businesses (SMBs) move to the cloud. For each one, you’ll see what it looks like on the ground and how to head it off before cutover weekend.
Short answer: The biggest cloud migration challenges for SMBs are legacy app compatibility, hidden dependencies, downtime during cutover, data transfer limits, unbudgeted costs, security gaps, missing skills, and user pushback. Nearly all of them trace back to an incomplete inventory before the move starts.
The migration itself is rarely the expensive part. The surprises are.
When McKinsey looked at where the overspend came from, the answer wasn’t cloud pricing. It was coordination: workloads moved without anyone assessing how complex they were, teams working out of sequence, and missing skills that nobody lined up in advance. Added up globally, the firm estimated those overruns would reach more than $100 billion in wasted migration spend over three years (McKinsey, 2021).
The bills don’t settle down after go-live, either. Flexera’s 2026 data puts wasted spend on infrastructure and platform services at an estimated 29%, the first increase in five years, and 17% of respondents went over their public cloud budgetin the past year (Flexera, 2026).
Here’s what the enterprise reports leave out. An SMB doesn’t have a Cloud Center of Excellence or a FinOps team. It has Dave. Dave also handles the phones, the firewall, and the Wi-Fi password everyone keeps forgetting, and when the plan has a gap, there’s no second person to catch it.
Most migration budgets don’t fail on pricing. They fail on the things nobody listed before the move.
The eight challenges below fall into two groups. The first four are technical. The second four are about money, security and people, and those are the ones business owners usually don’t see coming.
These are the ones that break things on day one.
Your accounting runs on QuickBooks Desktop or Sage 50. Operations lives in a Microsoft Access database someone built in 2011. And there’s a line-of-business app from a vendor that stopped answering emails years ago.
None of these are cloud-native. Most can’t move to a typical infrastructure as a service (IaaS) setup without real work, because they expect a Windows environment, a mapped drive pointing at a shared data file, and a user logged into a desktop. Rewriting them isn’t realistic for a 20-person firm. Replacing them means retraining everyone and converting years of data into a new format.
The fix: sort every application into one of three buckets before you choose a platform. Some can switch to a SaaS version (Microsoft 365 is the classic example). Some can be rehosted as-is on a Windows environment in the cloud. A few genuinely need replacing.
That middle bucket is almost always bigger than people expect. It’s also the one that decides what kind of cloud you actually need. At V2 Cloud, we run Windows apps like QuickBooks, Sage and AutoCAD this way through our application hosting service, with no code changes on the app side.
Ask anyone what the office server does. They’ll say “files and QuickBooks." Then you migrate it.
Dependencies are the quiet killer. Mapped drive letters hard-coded into Excel macros. Scan-to-folder settings on the copier. A SQL Server Express instance hiding behind the CRM. A USB license dongle. A scheduled task that emails the sales report every Friday. Each one works perfectly until the machine it relies on moves somewhere else.
The fix is boring, and it works. Log into the old server and list every installed program, running service, scheduled task, shared folder and printer. Then walk the office and ask each person what they open on a normal Monday and at month-end.
Twenty minutes per person beats two days of troubleshooting after cutover. If you haven’t built a plan yet, our cloud migration guide covers the assessment stage in more depth.
For a small business, “the system is down" usually means “nobody can invoice." A law office that loses its document system on a Tuesday afternoon loses billable hours it never gets back.
The common mistake is treating cutover as one big event. Run the old and new environments side by side for at least a few days.
Move a pilot group of two or three people first, ideally the ones who complain the loudest, since they’ll find problems fastest. Schedule the final data sync for a Friday evening. Then keep the old server powered on in read-only mode for about two weeks, in case something got missed.
A managed provider can shrink this window a lot. When businesses move off on-premise servers onto our platform, our customer success team handles the migration, and most are fully operational within a single business day. That’s a fair benchmark to ask any provider about.
Quick math. Moving 500 GB over an upload connection that really delivers 50 Mbps takes about 22 hours, assuming nothing else uses the line and nothing fails halfway through.
Most small offices have asymmetric internet plans, so real upload speed is a fraction of the headline number on the bill. And that 22 hours only covers the first copy. Files keep changing while the transfer runs.
The fix: clean up before you copy. Archive finished projects, delete duplicate backups, and move only what people actually use. Start the bulk transfer early, then run incremental syncs until cutover so the final copy is small. For very large data sets, ask your provider whether they accept a shipped drive.
Once you’re live, bandwidth matters in a different way. With a hosted desktop, the data stays in the data center and only screen updates travel to the user. A 2 GB project file opens at data-center speed, not home Wi-Fi speed.
Not sure whether your setup can move without a rebuild? Compare what a hosted setup would cost for your actual workloads.
These don’t crash anything on Monday. They show up later, in the invoice, the audit, or the resignation letter.
A server replacement quote is easy to compare against a monthly cloud bill. The line items that appear later are harder to spot: egress fees for pulling data back out, licensing that’s priced differently in the cloud, backup storage, extra consultant hours when plan A doesn’t work, and the management time someone still has to spend every month.
Usage-based billing adds another wrinkle. A virtual machine left running over a long weekend, or storage that quietly doubles, won’t show up anywhere until the invoice does.
The fix: build your estimate around three years of total cost of ownership, not the monthly sticker price. Ask every provider which costs are fixed and which are metered. Flat monthly pricing is easier to forecast than usage billing, even when the metered rate looks cheaper on paper. We broke down every cost driver in our guide to cloud migration costs.
Data in transit is exposed. So is data parked in a temporary share “just for the migration" that nobody ever deletes.
The migration window is often the least secure moment of the entire project. Admin accounts get broad permissions to make copying easier. Firewall rules get opened temporarily and stay open. A contractor gets a login that nobody remembers to disable.
The fix: encrypt data in transit and at rest, and turn on multi-factor authentication (MFA) for every account before any data moves, not after. Give migration accounts only the access they need, put an expiry date on contractor logins, and close temporary rules at cutover.
In a regulated industry, confirm in writing that the new environment meets your obligations (HIPAA for healthcare, for example), including a signed business associate agreement where one is required.
Your cloud provider secures the infrastructure. You still decide who gets a login.
Running a server in a closet and running infrastructure in AWS or Azure are different jobs. Networking works differently. So do identity, backups and cost tracking, and the learning curve lands on whoever is already stretched thinnest.
McKinsey’s outperformers, the companies that finished ahead of schedule, were far more likely to have hired staff with advanced DevOps and FinOps skills (McKinsey, 2021). Most SMBs can’t hire a DevOps engineer for one project.
The fix: decide honestly what you want to own afterward. If you want full control, budget for training or a certified partner. If you’d rather not become a part-time cloud administrator, pick a managed platform where patching, backups and monitoring are included, so your IT person manages users instead of infrastructure.
People don’t resist the cloud. They resist anything that makes Tuesday harder than Monday.
Ask staff to learn a new login, a new folder structure and a new way to print on the same day, and you’ll get pushback, workarounds, and client files saved to personal Dropbox accounts.
The fix: change as little as possible for end users in phase one. Keep the same Windows desktop, the same drive letters, the same apps. Tell people what’s changing and when, a week ahead, and give them one named person to call. Modernize workflows later, once the move itself has become boring.
The approach you choose decides which cloud migration challenges you’ll actually face. There are three common routes, and each trades one set of problems for another.
Lift-and-shift (rehosting) copies your servers into IaaS as virtual machines. It’s quick, but you bring every old problem along and take on the job of running cloud infrastructure.
Refactoring rebuilds applications on cloud-native services. It gives the best long-term fit and costs the most time and money. For off-the-shelf software like QuickBooks, it usually isn’t an option at all.
Desktop as a Service (DaaS) moves the whole Windows desktop, apps included, onto infrastructure the provider manages. Users log into the same environment from any device. Some people still call this hosted virtual desktop infrastructure (VDI), but the managed model removes the part of VDI that made it painful: building and maintaining the infrastructure yourself.
| Challenge | Lift-and-shift (IaaS) | Refactor (cloud-native) | DaaS (hosted desktops) |
|---|---|---|---|
| Legacy app compatibility | Apps run, but you manage the VMs | Apps must be rewritten or replaced | Apps run unchanged on Windows |
| Hidden dependencies | Must be mapped and rebuilt | Redesigned from scratch | Must be mapped; drives and printers recreated inside the desktop |
| Cutover downtime | Moderate | Spread over months of phased releases | Low; pilot users can move first |
| Cost predictability | Metered, harder to forecast | High upfront, variable after | Often flat monthly (confirm with provider) |
| Skills needed | Cloud administration | Developers plus cloud architects | Basic Windows administration |
| User disruption | Low | High (new tools and workflows) | Low (same desktop and apps) |
What this means: if most of your workloads are Windows desktop apps, a hosted desktop sidesteps most of the challenges above. If you’re building custom software, refactoring pays off over time.
You don’t have to pick just one. Gartner expects 90% of organizations to adopt a hybrid cloud approach through 2027(Gartner, 2024). For a small business, that often looks like Microsoft 365 as SaaS, legacy apps running on cloud desktops, and one on-site device for the thing that truly can’t move.
Our desktops deploy in under 20 minutes, which makes it practical to pilot your real apps before committing to anything.
Nearly every one of the cloud migration challenges above gets created during planning, even though it only shows up after cutover. This list covers what we’d check before touching a single file.
For a step-by-step walkthrough of the move itself, see our guide to on-premise to cloud migration.
One last thing. Keep the inventory you built in step 1. It’s the document every future IT person will wish existed, and it’s the first thing you’ll need if you ever switch providers again.
Want a second set of eyes on your migration plan before cutover weekend? Our team can walk through your apps and dependencies with you.
For most small and midsize businesses, it’s cost. Of all the cloud migration challenges, unbudgeted expenses such as egress fees, licensing changes and ongoing management do the most damage.
Moving a single office server can take days, while multi-site projects with custom applications can run for months. The timeline depends mostly on how many apps and dependencies you have. With our managed migrations, most businesses are fully operational within a single business day.
Yes. Most Windows applications can be rehosted unchanged on a cloud-hosted Windows desktop or server, including QuickBooks, Sage 50 and Microsoft Access. Check each vendor’s licensing terms first, since some restrict how and where their software can be hosted.
Run the old and new environments in parallel, move a small pilot group first, and schedule the final data sync outside business hours. Keep the old server available in read-only mode for about two weeks so anything missed can be recovered quickly.
Mostly because of costs left out of the original estimate: data transfer, licensing, extra consulting hours and ongoing management. Metered pricing makes it worse, since usage creeps up unnoticed. Estimating three-year total cost and choosing fixed monthly pricing where possible keeps forecasts realistic.
For businesses that run mostly Windows desktop apps, usually yes. Desktop as a Service keeps the desktop and apps users already know, and the provider manages the infrastructure, so there’s less to rebuild and less to learn. IaaS gives more control but requires cloud administration skills.
Your V2 CloudCare team — real people, on the line.
