Cloud Migration: The 90-Day Playbook

Most cloud migrations fail not because the technology is hard, but because the plan is wrong. Here is a 90-day framework that gets your first workloads into production while building the foundation for everything that follows.

Why 90 Days

Twelve-month migration plans fail. They lose executive sponsorship after quarter two. They accumulate scope creep that invalidates the original architecture. Key engineers leave. Business priorities shift. By month eight, the migration has become a zombie project: not dead enough to cancel, not alive enough to deliver.

Ninety days forces discipline. It demands prioritization. It requires a migration wave that delivers tangible value -- cost savings, performance improvement, or new capability -- fast enough to sustain momentum for the waves that follow.

This playbook is not about migrating your entire data center in 90 days. It is about establishing the cloud foundation and migrating your first wave of workloads in production within that window.

Days 1-30: Assessment & Foundation

Week 1-2: Portfolio Discovery

Inventory every application, database, and infrastructure component. For each, capture:

Business criticality -- What happens if this goes down for an hour? A day? A week?
Technical complexity -- Monolith or distributed? Stateful or stateless? What are the dependencies?
Data sensitivity -- PII, financial data, regulated data? What compliance requirements apply?
Migration disposition -- Rehost, replatform, refactor, rebuild, or retire?

Do not boil the ocean. A spreadsheet with 200 rows of accurate data is more useful than a CMDB with 2,000 rows of stale data. Focus on the applications that matter.

Week 2-3: Cloud Landing Zone

Build the foundational cloud infrastructure using Infrastructure as Code (Terraform or equivalent). This includes:

Account/subscription structure -- Multi-account topology aligned to your organizational model
Network topology -- VPCs, subnets, transit gateway, VPN or Direct Connect to on-premises
Identity foundation -- SSO integration, IAM roles and policies, service account governance
Security baseline -- GuardDuty/Defender, Config rules, CloudTrail/Activity Log, encryption policies
CI/CD pipeline -- Automated deployment pipeline for infrastructure changes

Every resource is defined in code. Nothing is created manually. This is non-negotiable.

Week 3-4: Wave 1 Selection

Select 3-5 workloads for the first migration wave based on:

Low risk, high visibility -- Applications that demonstrate cloud value without endangering critical operations
Simple dependencies -- Minimal integration with other on-premises systems
Clear success metrics -- Cost reduction, performance improvement, or feature enablement that is measurable

The goal is not to migrate the hardest thing first. It is to build confidence, prove the process, and create internal advocates.

Days 31-60: Migration Execution

Week 5-6: Environment Setup

For each Wave 1 workload, build the target environment:

Compute -- EC2/VM instances, container clusters, or serverless functions based on the application's profile
Data -- Database migration using DMS, native replication, or export/import based on size and downtime tolerance
Networking -- Load balancers, DNS, certificates, and connectivity to dependent services
Monitoring -- CloudWatch/Azure Monitor, application-level health checks, alerting pipelines

Week 7-8: Migration & Validation

Execute the migration for each workload:

Parallel run -- Deploy in cloud while keeping on-premises running. Route a percentage of traffic to cloud.
Validate -- Compare performance, data consistency, and functional behavior between environments.
Cutover -- Switch DNS/load balancer to cloud. Keep on-premises available for rollback.
Soak -- Run in production for 1-2 weeks with heightened monitoring before decommissioning on-premises.

Every migration has a documented rollback procedure. Every cutover has a pre-defined rollback trigger. No exceptions.

Days 61-90: Optimization & Wave 2 Planning

Week 9-10: Optimization

With Wave 1 in production, optimize:

Right-sizing -- Analyze actual utilization data and adjust instance sizes. Most workloads are over-provisioned by 40-60% in initial migration.
Reserved capacity -- Commit to reserved instances or savings plans for stable workloads
Architectural improvements -- Now that the workload is in cloud, identify opportunities to leverage cloud-native services (managed databases, object storage, serverless functions)

Week 11-12: Wave 2 Planning

Apply lessons from Wave 1 to plan Wave 2:

What took longer than expected? (Usually: data migration, network configuration, and identity/access setup)
What went smoother than expected? (Usually: compute migration, monitoring setup)
What should Wave 2 include? (Typically: more complex workloads, more dependencies, potentially stateful applications)

Document the playbook refinements. Update the Infrastructure as Code modules. Train additional team members on the process.

Common Mistakes to Avoid

Lift-and-shift everything. Some applications should be replatformed or rebuilt. If you are moving a monolithic Java application to EC2, you are paying cloud prices for on-premises architecture. Evaluate each workload honestly.

Skip the landing zone. Building infrastructure manually "to save time" creates security gaps, inconsistency, and technical debt that slows every subsequent wave. The landing zone investment pays for itself within the first wave.

Underestimate data migration. Database migration is almost always the critical path. Start data replication early, test it thoroughly, and have a fallback plan for every data store.

Ignore cost management from day one. Cloud cost surprises kill executive support faster than technical problems. Implement tagging, budgets, and cost alerts before you migrate a single workload.

Neglect the team. Cloud migration is a skills transformation as much as a technology transformation. Invest in training, pair on-premises engineers with cloud-experienced architects, and build internal capability from day one.

What Success Looks Like at Day 90

Cloud landing zone deployed and operational with IaC, monitoring, and security controls
3-5 workloads running in production on cloud infrastructure
Measurable cost savings or performance improvements documented
Migration playbook refined based on Wave 1 experience
Wave 2 plan approved and scheduled
Internal team trained and capable of executing subsequent waves with decreasing external support

The 90-day playbook is not the end of the migration. It is the beginning of a multi-year journey. But it is the phase that determines whether that journey delivers value or becomes a cautionary tale.

David Boggess is Managing Partner and CEO of Boggess Advisory Group. BAG has led cloud migrations for organizations managing billions in assets, achieving 90% data center reduction and $6M+ annual savings.