Lift-and-shift vs Replatform vs Refactor
The strategy you pick changes your total bill 3–5×. Here is what each approach actually costs per workload, by workload type, with the 2026 industry rates underneath.
Three strategies, head-to-head
| Metric | Rehost | Replatform | Refactor |
|---|---|---|---|
| Upfront / workload | $3K–$8K | $8K–$25K | $25K–$80K |
| Timeline / workload | 2–4 wk | 4–12 wk | 12–52 wk |
| Post-migration cloud cost | 100% (baseline) | ~85% | ~60% |
| 3-year TCO Δ vs on-prem | +10% | −15% | −40% |
| Risk | Low | Medium | High |
| Best for | Legacy apps, tight timelines, early cloud adoption | Databases, mid-life apps, teams with some cloud skills | Strategic apps, microservices targets, high-traffic consumer apps |
Cost by workload type
Different application types have very different per-workload migration costs. This is the table every vendor blog describes in prose.
| Workload | Rehost | Replatform | Refactor |
|---|---|---|---|
| Web Application | $2K–$5K | $5K–$15K | $20K–$60K |
| SQL Database | $3K–$10K | $10K–$30K | $30K–$100K |
| Batch Processing Job | $1K–$3K | $3K–$10K | $15K–$50K |
| Data Warehouse | $8K–$25K | $20K–$60K | $60K–$200K |
| Monolithic Application | $5K–$15K | $15K–$40K | $50K–$200K |
Per-workload cost (not per-server). 2026 industry consulting rates ($150–$350/hr). Varies by provider, region, and complexity.
The 7Rs migration framework
Rehost
Lift-and-shift. Move as-is to cloud.
- Cost
- $3K–$8K / workload
- Time
- Weeks
Replatform
Targeted cloud optimisation without full re-architecture.
- Cost
- $8K–$25K / workload
- Time
- 1–3 mo
Refactor
Rebuild for cloud-native architecture.
- Cost
- $25K–$80K+ / workload
- Time
- 3–12 mo
Repurchase
Move from on-prem to SaaS (e.g. CRM to Salesforce).
- Cost
- SaaS subscription
- Time
- 1–4 wk
Retire
Decommission applications no longer needed.
- Cost
- $0 (saves cost)
- Time
- Immediate
Retain
Keep on-premise. Too complex, regulated, or costly to move.
- Cost
- $0 migration
- Time
- n/a
Relocate
Move to cloud infra layer without changing OS.
- Cost
- $1K–$3K / workload
- Time
- Days
What strategy mix should you use?
Industry data from Gartner and McKinsey on what enterprise migration programmes actually look like in practice.
Legacy apps, tight timelines, retire-soon workloads
Databases, mid-life apps
Strategic apps, consumer-facing, high-traffic
Unused or redundant apps found in discovery
Decision framework
Classify each workload by strategic importance × modernisation benefit × urgency. High strategic + high modernisation = refactor. Low strategic + tight timeline = rehost. Everything in between = replatform.
Strategy questions
Q. What is lift-and-shift migration and how much does it cost?
Lift-and-shift (rehost) moves workloads to cloud with minimal modification. It costs $3,000-$8,000 per workload upfront, the cheapest migration strategy. Because workloads are not optimised for cloud, ongoing costs run 5-15% higher than on-premise over 3 years. Best for: legacy apps with tight timelines, early cloud adoption, workloads planned for retirement within 2 years.
Q. What is the difference between replatform and refactor cost?
Replatform costs $8,000-$25,000 per workload and reduces ongoing cloud costs 10-20% vs lift-and-shift. Refactor costs $25,000-$80,000+ per workload but reduces ongoing cloud costs 30-50%. For a 100-workload migration: replatform adds $800K-$2.5M vs lift-and-shift's $300K-$800K, but typically saves $60K-$120K per year on the run.
Q. What is the 7Rs cloud migration framework?
The 7Rs are: Rehost (lift-and-shift), Replatform (targeted optimisation), Refactor / Re-architect, Repurchase (move to SaaS), Retire (decommission), Retain (keep on-premise), Relocate (cloud infrastructure layer without OS changes). The most commonly used in enterprise migrations are Rehost (40-60% of workloads), Replatform (20-30%), and Retire (10-15%).
Q. What is the cheapest way to migrate to cloud?
Lift-and-shift is cheapest upfront. To minimise total cost: use free migration tools (AWS Application Migration Service, Azure Migrate, Google Cloud Migrate), apply for partner credits (AWS MAP, Azure FastTrack), right-size before reserving capacity, retire workloads before migrating, and ship large datasets physically (AWS Snowball) to avoid egress fees.