Most cloud migrations that go wrong, go wrong in the same few ways. A project is budgeted on a guess, the timeline assumes everything will copy across cleanly, nobody tests from where users actually sit, security is left on default settings, and the people who use the systems hear about it late.

None of this is exotic. It is a short list of avoidable mistakes, and each one has an early warning sign.

This guide covers five of them, the warning signs to watch for, and what to do instead. The examples are illustrative patterns, not client stories.


Mistake #1: Migrating Without Understanding What You Actually Have

What it looks like:

A company decides to “move everything to Azure”. They have a rough inventory: some servers, some databases, a bunch of applications.

The migration starts. Partway through, the team discovers:

  • Undocumented databases feeding critical reports
  • A legacy application still in daily use that nobody mentioned
  • A custom integration between two systems that is on no documentation
  • Shared drives holding far more data than estimated

The project stops while someone documents everything, and the budget and timeline no longer match reality.

Why it happens:

IT teams know what’s documented. They don’t always know what’s actually running.

Applications accumulate over years. Someone builds a quick Access database. Marketing runs a FileMaker system. Operations has a spreadsheet macro that is somehow business-critical.

These systems aren’t in the IT asset register. They just exist, until migration breaks them.

What good discovery involves:

The point of discovery is to find what the documents don’t tell you:

  • Interview people across all departments, not only IT
  • Monitor network traffic to find undocumented systems
  • Review user folders on file servers
  • Check scheduled tasks and cron jobs
  • Look at database connections to identify hidden systems

Discovery routinely turns up systems IT didn’t know about, integrations built by people who have since left, and “decommissioned” systems that are still in use. Finding them before the migration is far cheaper than finding them when users complain.

How to avoid this mistake:

Step 1: Technical Discovery

  • Network traffic analysis (find all connections)
  • Database dependency mapping (what talks to what)
  • Server inventory (everything running, not just what’s documented)
  • Application dependencies (list every service, every integration)
  • Storage analysis (actual sizes, not estimated)

Step 2: Business Process Discovery

  • Interview department heads (what systems do they use daily)
  • Survey end users (what breaks if X stops working)
  • Review business processes (what’s actually business-critical)
  • Document peak usage times (when can’t you afford downtime)

Step 3: Create Migration Inventory

  • Everything gets categorized: Business Critical / Important / Nice to Have
  • Document all dependencies
  • Identify what can’t move (legacy systems, compliance restrictions)
  • Calculate actual data volumes
  • Map out integration points

Skipping discovery is the most common cause of budget and timeline overruns. Treat it as part of the project, not an optional extra. For how discovery fits into the rest of the schedule, see what to expect from a cloud migration timeline.

Cloud Migration Discovery Checklist

Mistake #2: Underestimating Data Migration Complexity and Time

What it looks like:

A firm has a large file server. “We’ll just copy it over the weekend.”

The weekend comes and the copy starts on Friday evening. By Monday morning, only part of it has transferred, files are corrupting, and production is delayed. They have to roll back and start again with a proper migration approach.

The problem most miss:

Data migration isn’t just copying files. It’s:

  • Maintaining file permissions and ownership
  • Preserving timestamps and metadata
  • Handling locked files and open connections
  • Dealing with long file paths (Windows has a 260 character limit)
  • Managing network bandwidth (don’t saturate your internet during business hours)
  • Validating data integrity after transfer
  • Cleaning up data (duplicate files, old versions)

Why estimates fail:

  1. Bandwidth limitations

    • Your upload speed on paper is not what you get in practice
    • Other traffic competes during business hours
    • Arithmetic helps: as an illustration, 1TB over a sustained 60Mbps link is about 37 hours of upload alone, before any processing or retries
  2. File system limitations

    • Windows 260 character path limit
    • Special characters in filenames
    • Case-sensitive vs case-insensitive file systems
    • NTFS permissions don’t map directly to cloud storage ACLs
  3. Application data complexity

    • Databases can’t just be copied (backups need restoring and validating)
    • Active Directory integrations need reconfiguration
    • Encrypted files need keys migrated
    • Applications with embedded file paths break

For very large data sets, copying over the internet may be the wrong approach altogether. Both clouds offer physical transfer devices: AWS Snowball and Azure Data Box.

How to avoid this mistake:

Phase 1: Data Assessment (Before Estimating)

  • Actual data volumes (not “approximately”)
  • File type breakdown (millions of small files vs. few large files)
  • Permission complexity (simple vs. complex ACLs)
  • Data cleanup potential (duplicates, old versions)
  • Network bandwidth available (measured, not theoretical)

Phase 2: Test Migration

  • Migrate a representative sample first
  • Measure actual transfer rates
  • Identify issues (long paths, permissions, corrupted files)
  • Calculate a realistic timeline from the test results

Phase 3: Data Cleanup (Before Migration)

  • Remove duplicates (file hash comparison)
  • Archive old data that is no longer accessed
  • Compress where appropriate
  • Fix permission issues before migrating

Phase 4: Staged Migration

  • Start with non-critical data
  • Migrate in chunks, validate each chunk
  • Don’t migrate everything at once
  • Keep source data until validation is complete

Cleaning up before you move is the cheapest saving in the whole project. You don’t pay to transfer, store or back up data you didn’t need.


Mistake #3: Ignoring Network Connectivity and Latency

What it looks like:

A company migrates an application to a cloud region. It works fine in testing.

Launch day: users in the office say the application is slow, and remote workers say it is unusable.

The problem? They tested from inside the cloud, where everything is fast. They didn’t test over actual user connections.

What went wrong:

On-premise, the application and database sat on the local network, so latency was tiny and page loads were quick.

After migration, every user is now a network hop away from the application, and users in other cities are further away than those in the head office. A step that was imperceptible on a local network is noticeable at that distance.

Why latency matters more than you think:

Traditional client-server applications make a lot of round trips:

  • Load page, query database (1 round trip)
  • Click dropdown, query for options (1 round trip)
  • Fill form, validate each field (several round trips)
  • Submit, process and redirect (a few more)

The rule of thumb is: latency impact = number of round trips x latency per round trip.

On a local network, each round trip is close to instant, so even many of them are barely noticeable. Across a wide area network, the same number of round trips adds up. A chatty application that made dozens of database calls per page load will feel slow from a distant city, even though nothing is wrong with the cloud.

Typical fixes are to reduce the number of round trips, add a caching layer, or place components in a region closer to users. You can measure latency from each office to each region with simple tools, so there is no excuse for guessing.

Network connectivity issues you’ll hit:

1. VPN Performance

  • Site-to-site VPN throughput is limited
  • VPN adds latency
  • Multiple offices mean multiple VPN tunnels to manage

2. Direct Connect / ExpressRoute Costs

  • A dedicated connection costs a circuit fee plus data transfer
  • Check current prices on the AWS Direct Connect and Azure ExpressRoute pages
  • Dedicated connections also take time to provision

3. Hybrid Scenarios

  • Some systems stay on-premise
  • Some move to cloud
  • Constant data sync between them
  • Network becomes the bottleneck

4. International Connectivity

  • Data stays in Australia for sovereignty
  • International users (if you have them) get poor performance
  • A CDN is usually required for an acceptable experience

How to avoid this mistake:

Before Migration:

1. Test From Actual User Locations

  • Not from the data centre
  • Not from your desk in head office
  • From branch offices, remote workers, wherever users actually are

2. Measure Current Performance Baseline

  • Application page load times
  • Database query response times
  • File access speeds
  • Peak usage periods

3. Model Cloud Performance

  • Test latency from each location to the candidate cloud regions
  • Estimate the expected performance change
  • Decide whether it is acceptable or needs optimization

4. Design for Latency

  • Reduce database round trips
  • Implement caching where appropriate
  • Use a CDN for static content
  • Consider edge options for remote locations

Network Design Options:

Option 1: Public Internet Only

  • Pros: Cheapest, easiest
  • Cons: Variable performance, security concerns, latency
  • Good for: Non-critical apps, small teams, low data volumes

Option 2: Site-to-Site VPN

  • Pros: Secure, affordable
  • Cons: Throughput limited, latency overhead
  • Good for: SMEs with a few offices and moderate data transfer

Option 3: Direct Connect / ExpressRoute

  • Pros: Fast, reliable, predictable performance
  • Cons: Expensive, takes time to provision
  • Good for: Larger organisations, high data volumes, latency-sensitive apps

Option 4: Hybrid Approach

  • VPN for general access
  • Dedicated connection for critical applications
  • CDN for public-facing content

Choose the simplest option that meets your performance needs, and build it into the original design. Retrofitting connectivity after launch means a long stretch of user complaints.

Mistake #4: Underestimating Security and Compliance Requirements

What it looks like:

A healthcare provider migrates a patient management system to the cloud. Some time later, a compliance audit finds:

  • Data stored in an overseas region
  • Audit logging not configured
  • Backup encryption not enabled
  • Multi-factor authentication not enforced

That is a compliance failure. The fix is to migrate again into an Australian region, implement proper logging, enable encryption, set up MFA and commission an external security audit. All of it costs more than doing it right the first time.

Common security mistakes:

1. Wrong Region Selection

Many Australian businesses have obligations that make data location matter:

  • Health data: Australian Privacy Principles and health privacy laws
  • Financial services: APRA requirements
  • Government contractors: IRAP-assessed services

Depending on your obligations, you may need to keep data in Australia, or at least know and account for exactly where it goes. Check your own requirements, and get advice if you’re unsure.

It still happens that data ends up in an overseas region because it was cheaper, had more services or was faster to provision. Fixing that later is expensive, slow and a compliance risk. Our data sovereignty and compliance guide sets out what to check before you choose a region.

2. Default Security Settings

Cloud providers optimize for ease of use, not security.

Defaults that bite you:

  • Storage buckets accessible more widely than intended (hello data breaches)
  • Weak password policies (no MFA requirement)
  • Broad network access rules (open to the internet)
  • Minimal logging (can’t prove compliance)
  • Encryption optional (not always on by default)

A typical audit of a rushed deployment finds public storage, admin accounts without MFA, databases accepting connections from any address, no encryption at rest and minimal audit logging. None of it is malicious. Nobody changed the defaults.

3. Inadequate Access Controls

On-premise mindset: “Everyone’s on our network, so it’s secure.”

Cloud reality: “Internet-accessible, so it needs proper controls.”

Common issues:

  • Shared admin passwords
  • No role-based access control
  • Former employees still have access
  • Third-party contractor access not reviewed
  • No multi-factor authentication

A common finding after migration is a former employee whose access was never revoked in the new cloud environment. Account clean-up should be a named task in the migration plan.

4. Backup and Disaster Recovery Assumptions

“It’s in the cloud, they back it up, right?”

No. Cloud providers guarantee infrastructure availability. They don’t guarantee YOUR data protection.

You’re responsible for:

  • Database backups
  • Application data backups
  • Configuration backups
  • Testing restore procedures
  • Disaster recovery planning

Data loss from an application bug or accidental deletion is not the provider’s fault, and without backups you can’t recover it. The worst case is losing recent transactions that then have to be rebuilt from emails and paper records.

How to avoid this mistake:

Before Migration - Compliance Assessment:

1. Identify Requirements

  • Industry regulations (healthcare, financial services, government)
  • Data residency requirements
  • Audit and logging requirements
  • Encryption requirements
  • Access control requirements

2. Document Controls

  • What security controls exist today
  • What the cloud provider offers natively
  • What gaps need addressing
  • What additional tools or services are needed

3. Design Security Architecture

  • Network segmentation
  • Identity and access management
  • Encryption (at rest and in transit)
  • Logging and monitoring
  • Backup and disaster recovery

Security Checklist for Australian Businesses:

Data Sovereignty ✓

  • All data in Australian regions (Sydney and/or Melbourne), where required
  • Verify backup locations (including snapshots, replicas)
  • Document data flows (especially international transfers)
  • Configure geo-restrictions where required

Encryption ✓

  • Encryption at rest for all storage
  • Encryption in transit (TLS 1.2 minimum)
  • Key management strategy
  • Database encryption enabled

Access Control ✓

  • Multi-factor authentication enforced
  • Role-based access control configured
  • Privileged access management
  • Regular access reviews
  • Former employee access revoked

Monitoring & Logging ✓

  • Centralized logging configured
  • Audit trails for all admin actions
  • Security monitoring and alerting
  • Log retention meets compliance requirements

Backup & DR ✓

  • Automated backup configured
  • Regular backup restore tests
  • Disaster recovery plan documented
  • RTO and RPO defined and tested

Network Security ✓

  • Network segmentation implemented
  • Firewall rules configured (least privilege)
  • VPN or private connectivity
  • DDoS protection enabled

What to budget for:

Security has a one-time cost (design and setup), an ongoing cost (monitoring and management), and for regulated businesses an audit cost. How much depends on how regulated you are and how much you run. Price each from your own requirements, and don’t leave it as a gap in the migration budget.

The cost of non-compliance is harder to put a number on: a data breach, regulatory consequences and reputation damage. For regulated businesses, that is the strongest argument for getting this right up front.


Mistake #5: Poor Change Management and User Adoption

What it looks like:

A company completes a technically perfect migration. Zero downtime, all systems working.

Then adoption stalls. Many users are still trying to use the old system, which has been turned off. Productivity drops.

The problem:

Technical success is not the same as business success.

Users need to:

  • Understand why change is happening
  • Know how to use the new systems
  • Have support when things go wrong
  • Feel involved in the process

What goes wrong:

Management announces the migration at the last minute. There is no training and no documentation, on the assumption that “they’ll figure it out.”

They don’t. Instead:

  • They call IT constantly
  • They make errors in the new system
  • They blame the migration for everything
  • Some find workarounds, such as emailing spreadsheets instead of using the system

Change management mistakes:

1. Treating It as an IT Project (Not a Business Project)

Technical teams focus on servers, databases and uptime.

Users care about: “Can I do my job?”

A migration can be declared a success by IT (all systems migrated) while staff can’t find documents because of a new folder structure, can’t generate invoices because report paths changed, and client communications are delayed.

IT: “Everything’s working.” Business: “We can’t work.”

Both are correct. They have different definitions of “working”.

2. Insufficient Training

“It’s the same system, just in the cloud” is famously wrong.

What changes for users:

  • Login process (now web-based instead of domain login)
  • VPN required for some access
  • File paths different (mapped drives become OneDrive/SharePoint)
  • Printers need reconfiguring
  • Saved shortcuts don’t work
  • Application performance might feel different

What works:

  • Training sessions for all users, during work time
  • Quick reference guides
  • Champions in each department (super users who get extra training)
  • Extended IT support at the start
  • Regular check-ins in the first week

3. No Support Plan for Post-Migration

“Go live” isn’t the end. It’s the beginning of user challenges.

Typical issues in the first week after migration:

  • “How do I access X?”
  • “Where did my files go?”
  • “This is slower than before”
  • “I can’t print”
  • “My shortcuts don’t work”

If you don’t have a support plan, these small issues compound. The IT team goes back to “normal” work, tickets pile up, issues aren’t resolved quickly, frustration builds, and the perception becomes “the migration failed”.

Technical reality: everything is working fine. User reality: can’t get help when needed.

4. All-or-Nothing Approach

The biggest risk is migrating everything at once. Practice management, email, file servers and line-of-business apps all change the same weekend. On Monday morning everything is unfamiliar at once, and both users and the support team are overwhelmed.

A better approach is to phase it:

  • File servers first
  • Email next, once users are comfortable with the first change
  • Line-of-business applications after that
  • Remaining systems last

Each phase means users learn one change at a time, IT focuses support on one area, issues are found and fixed before the next phase, and confidence builds.

How to avoid this mistake:

Change Management Plan:

Phase 1: Communication (well before migration)

Who, what, why, when:

  • What’s changing
  • Why we’re doing this
  • Benefits (faster, more reliable, cost savings, whatever’s true)
  • Timeline
  • What users need to do
  • Where to get help

Ongoing communication:

  • Regular updates
  • Progress reports
  • Address concerns
  • Celebrate milestones

Phase 2: Training (ahead of migration)

Training program:

  • Role-based training (different roles need different focus)
  • Hands-on practice (not just slides)
  • Quick reference guides (one-page cheat sheets)
  • Video tutorials (for self-service learning)
  • Super user program (champions in each department)
  • A lab environment to practise in, if possible

Phase 3: Phased Rollout (Minimize risk)

Approach:

  1. Pilot group (IT team + volunteers)
  2. Department by department
  3. Critical systems last (when confident in the process)

Each phase:

  • Test before rolling out to more users
  • Gather feedback
  • Adjust approach
  • Fix issues
  • Document lessons learned

Phase 4: Support (After migration)

Support strategy:

  • Extended IT hours, especially in the first couple of weeks
  • A dedicated migration support channel
  • Daily stand-ups to address issues
  • Quick response to problems
  • Regular user feedback sessions

Budget for change management, training development and extended support as line items in the project. They are cheap compared with a migration that people don’t adopt.


Pulling It All Together: What Good Migration Looks Like

A well-run migration follows a predictable shape:

Phase 1: Discovery

  • Document all systems
  • Interview users
  • Identify dependencies
  • Find systems IT didn’t know about
  • Create an accurate inventory

Phase 2: Planning

  • Design the cloud architecture
  • Plan network connectivity
  • Address security requirements
  • Create the migration schedule
  • Budget accurately

Phase 3: Testing

  • Migrate pilot systems
  • Test performance from all locations
  • Validate security controls
  • Refine the approach based on what you learn

Phase 4: User Preparation

  • Communicate changes
  • Conduct training
  • Create reference materials
  • Build confidence

Phase 5: Migration (phased)

  • Non-critical systems first
  • File servers and email next
  • Business applications after that

Phase 6: Stabilization

  • Extended support
  • Issue resolution
  • Optimization
  • User feedback

The difference:

  • Proper discovery
  • Realistic planning
  • Phased approach
  • Strong change management
  • Adequate support

Summary: How to Avoid These Mistakes

Mistake #1: Incomplete Discovery

  • Invest properly in discovery
  • Interview users, not just IT
  • Document everything actually running
  • Test assumptions

Mistake #2: Underestimating Data Migration

  • Measure actual data volumes
  • Test the migration approach first
  • Clean data before migrating
  • Plan for bandwidth limitations
  • Validate after migration

Mistake #3: Ignoring Network Performance

  • Test from actual user locations
  • Measure latency impact
  • Design for latency (caching, optimization)
  • Consider Direct Connect/ExpressRoute for critical apps

Mistake #4: Inadequate Security

  • Use Australian regions where your obligations require it
  • Don’t accept default security settings
  • Implement proper access controls
  • Configure encryption and logging
  • Test backup and recovery

Mistake #5: Poor Change Management

  • Communicate early and often
  • Train users properly
  • Phased rollout (not all at once)
  • Extended support post-migration

Doing this properly costs time and money up front, and that investment goes into discovery, planning, security and change management. Skipping those steps to save on the project usually costs more later in overruns, rework, poor adoption and compliance problems.

If you want help planning your migration, get in touch.