2026 Playbook

Cloud Migration: The 2026 Mid-Market CIO's Playbook

A 5-axis Readiness Score, the 7Rs as a decision sequence, the agentic-AI tooling shift, and 17 consulting firms evaluated on platform depth, case-study evidence, and pricing transparency. Built for IT leaders running 50–500 application portfolios on a 9–18 month window.

By Peter Korpak, Founder · Last updated: June 9, 2026 · See our methodology

The 2026 Reality

Most cloud migrations don't deliver. Three numbers tell you why.

The hyperscaler "what is cloud migration" pages won't print these. We will. Read them, then keep going — the rest of this page is about how mid-market CIOs avoid being in any of these statistics.

55%

of cloud migrations failed to meet expectations; 53% didn't deliver promised benefits.

Gitnux 2026

29%

of cloud spend is wasted in 2026, reversing a five-year decline. The top wasted line is over-provisioned compute and unrightsized DBs.

Flexera 2026 State of the Cloud

93%

of enterprises have repatriated AI workloads, are doing so, or are evaluating it. The repatriation conversation is now table stakes for any AI-adjacent migration.

Cloudian Enterprise AI Infrastructure Survey 2026 (n=203)

The cloud migration services market is worth roughly $330B in 2026, growing at a 23.2% CAGR (Research and Markets, Feb 2026). Inside that, the mid-market segment — 500 to 5,000 employees — is the fastest-growing slice at 17.65% annually per Auvik 2026. The Fortune 500 playbooks built for $5M–$20M migrations don't translate down. Most of the budget overruns and 55% miss-rate sit in the segment that's growing fastest — and getting the worst guidance.

The single biggest unforced error in the past three years: picking a strategy before scoring readiness. Vendor frameworks (AWS MRA, Azure Strategic Migration Assessment) push 50–100 question maturity audits that produce a red/yellow/green level. They don't tell you which R to pick. We built one that does.

Note for 2026 planning: AWS Migration Hub closed to new customers on November 7, 2025, replaced by AWS Transform (agentic AI). Azure released its Copilot Migration Agent the same month. Google's Migration Center got a Gemini-powered App Modernization Assessment in September 2025. If your prospective consulting partner is still recommending Migration Hub for new engagements, you're talking to a firm whose information is six months stale.

Framework

The 7Rs as a decision sequence (not a menu)

Hyperscaler pages list the 7Rs as equal options. They're not. They're a funnel — a series of yes/no decisions you apply in order to each application. Below: the decision flow, with portfolio percentages from a typical mid-market 100-app shop.

The 7Rs as a decision sequence for cloud migration A horizontal decision flow diagram. Each application enters at the left and runs through five sequential questions: kill it (Retire), replace with SaaS (Repurchase), must stay on-prem (Retain), time pressure (Rehost), and modernization needed (Replatform / Refactor / Rearchitect). Portfolio percentages shown on each branch reflect a typical mid-market 100-app portfolio. Application Portfolio entry Still in use? Real users last 90d no Retire ~20% of portfolio yes Mature SaaS replacement? yes Repurchase ~15% of portfolio no Must stay on-prem? yes Retain ~10% of portfolio no Time-to-cutover < 6 months? yes Rehost ~30% of portfolio no Cloud-native benefit ROI? modest Replatform / Relocate ~20% of portfolio high Refactor / Rearchitect ~5% of portfolio Portfolio percentages reflect a typical mid-market 100-app discovery; your distribution will vary by ops maturity and regulatory load.

The shift this diagram makes: the first three Rs (Retire, Repurchase, Retain) shrink your migration scope before you migrate anything. AWS, Azure, and Google all show the 7Rs as a side-by-side menu. We treat them as a sequence — apply each filter, count what's left, and then plan capacity. A typical mid-market discovery returns ~20% of the portfolio retire-able, ~15% replaceable with a SaaS already on the market. Skipping these gates is how 50-app projects become 80-app projects.

Need the deep dive? Read Legacy System Modernization Approaches for the 7Rs at the workload level, or Cloud Migration Assessment Checklist for the discovery step that feeds this filter.

Cloud Intel Original

The Migration Readiness Score: 5 axes, one strategy

AWS Migration Readiness Assessment runs roughly 70 questions and outputs a maturity color. Azure Strategic Migration Assessment is similar. Both produce a level. Neither tells you which R to pick. The Cloud Intel Migration Readiness Score is a 5-axis vector that maps directly to a primary R, a secondary R, and — when ops maturity is too low — an honest "delay" output most vendor frameworks won't print.

Migration Readiness Score: pentagon radar with two profile overlays A pentagon radar chart with five axes — Data Sensitivity, Vendor Lock-in Tolerance, Ops Maturity, Regulatory Load, and Time-to-Cutover Pressure — each scored 1 to 5. Two profiles overlay: a mid-market SaaS company with low scores across the board, and a regulated fintech with high scores on Data Sensitivity, Regulatory Load, and Time-to-Cutover Pressure. Each profile maps to a different recommended migration strategy. Data Sensitivity Lock-in Tolerance Ops Maturity Regulatory Load Time-to- Cutover 5 4 3 2 1 Mid-market SaaS, no PII (low across the board) Regulated fintech, hard deadline (high on 4 of 5 axes) Profile A — Mid-market SaaS Sensitivity 2 · Lock-in 2 · Ops 4 · Reg 1 · Time 2 Primary R Refactor Secondary R Replatform Avoid Big-bang Rehost Time-to-cutover 9–14 months Partner type Cloud-native specialist Profile B — Regulated fintech Sensitivity 5 · Lock-in 4 · Ops 3 · Reg 5 · Time 5 Primary R Hybrid: Replatform + Retain (regulated) Secondary R Selective Refactor Avoid Refactor-everything Time-to-cutover 5–7 months (compressed) Partner type FinServ + sovereign region SI Key risk SoD & data residency drift

The five axes

Score each 1 (low) to 5 (high). Each axis controls a different dimension of the strategy decision.

Axis 1 (low) 3 (mid) 5 (high) What it controls
Data sensitivity Public / marketing PII, SOC 2 PHI, payment, classified, EU-regulated Filters out pure public cloud at high scores; pushes toward private region or sovereign cloud
Vendor lock-in tolerance Greenfield, hyperscaler PaaS OK Mixed; managed DBs OK Multi-cloud mandate, OSS-only, exit clauses High → portable runtimes (containers, OSS DBs), Replatform > Repurchase
Ops maturity No CI/CD, manual change mgmt Some IaC, basic monitoring Full IaC, SRE, DORA-elite Low → Rehost (don't refactor what you can't operate); high → Refactor viable
Regulatory load B2C SaaS, none One framework (SOC 2, ISO 27001) HIPAA + PCI + GDPR + sector (DORA, FedRAMP) High → wave plan with compliance gates, possibly Retain or repatriate
Time-to-cutover budget 24+ months acceptable 9–18 months <6 months (lease, M&A) High pressure forces Rehost first; low pressure makes Refactor cost-effective

Mapping your score to a strategy

Read the score as a vector, not a sum. The combination of axes maps to a primary R, a secondary R, and — for low ops maturity with high regulatory load — an honest delay output.

Score pattern Primary R Secondary R Why
Sensitivity ≤ 2, Lock-in ≤ 2, Ops 3–5, Time 1–2 Refactor Replatform Greenfield-ish posture, room to do it right
Sensitivity 4–5, Lock-in 4–5, Ops 4–5, Reg 4–5 Replatform (private/sovereign) Retain (regulated subset) Portability + control mandate; private region or sovereign cloud
Time = 5 (hard deadline), Ops ≤ 2 Rehost Retire (kill 25%+) Speed forces lift-and-shift; aggressive retire to shrink scope
Sensitivity 4–5, Reg 4–5, Time 1–3 Hybrid (Replatform + Retain) Selective Refactor Compliance-led — keep regulated workloads private
AI-heavy + Sensitivity 4–5 Repatriation-aware Hybrid Replatform burst tier Cloudian 2026: 91% of regulated AI workloads landing on-prem or private
Lock-in = 5, Ops = 5 Repurchase + Refactor Retire Mature team, lock-in averse — split portfolio: SaaS for commodity, OSS containers for differentiator
All axes ≤ 2 Rehost Retire Not ready for more — get to cloud, optimize later
Ops ≤ 2 + Reg ≥ 4 DELAY Address ops gap first Migrating regulated workloads with low ops maturity is how breaches happen — see Risk Assessment spoke

The "DELAY" output is the differentiator. Vendor frameworks always recommend "yes, migrate now" — that's how the SOWs get signed. We tell you when to wait. If you scored Ops ≤ 2 with high regulatory load, read Cloud Migration Risk Assessment and Cloud Migration Best Practices before signing any SOW.

Worked Examples

Three mid-market profiles, three different strategies

Each profile is composite — built from the engagement patterns we see most often in our directory's discovery interviews. The score values reflect what discovery typically returns, not what teams self-report.

Profile 1

Mid-market healthcare SaaS

~120 apps · 14-month window · HIPAA + SOC 2 · 85 employees, growing

Score: Sensitivity 4 · Lock-in 3 · Ops 3 · Reg 4 · Time 3

Strategy: Hybrid Replatform + selective Retain. Containerize non-PHI services on AWS or GCP; keep PHI database on dedicated tenancy with BAA-eligible services only. Refactor the patient-portal stack post-migration, not during.

Common mistake: Picking AWS GovCloud "to be safe" — overkill and expensive for non-Federal HIPAA. Standard AWS or GCP HIPAA-eligible regions are the right answer.

Partner type: Healthcare-vertical specialist with proven BAA scope on the target hyperscaler.

Profile 2

Retail with M&A deadline

~80 apps · 5-month window · Datacenter lease ends · Recent acquisition

Score: Sensitivity 2 · Lock-in 2 · Ops 2 · Reg 2 · Time 5

Strategy: Rehost first, retire aggressively. Lift-and-shift the 60% of apps that are still in real use; retire the rest. Replatform critical billing systems in wave 2 (post-cutover). Refactor nothing in this window.

Common mistake: Trying to refactor under a 5-month deadline. Mid-market teams burn the budget on the first 5 services and miss the lease.

Partner type: Migration factory with automated discovery (AWS Transform, Azure Migrate Copilot, or proprietary tooling). Avoid bespoke shops — they'll over-engineer.

Profile 3

Regulated fintech, hard deadline

~200 apps · 7-month window · DORA + FFIEC + SOX · EU + US

Score: Sensitivity 5 · Lock-in 4 · Ops 3 · Reg 5 · Time 5

Strategy: Hybrid Replatform with sovereign-region wave plan. Retain the core banking ledger on-prem in year 1; replatform peripheral services to EU sovereign region first, US workloads in wave 2. Refactor zero in this window.

Common mistake: Lifting the ledger onto a hyperscaler "to look modern." Regulators care about residency and audit trails, not how containerized your stack is.

Partner type: Financial-services SI with prior DORA work and explicit sovereign-region capacity. Not a generalist factory.

2026 Tooling Shift

Agentic AI changed migration economics in the last six months

From late 2025 through Q1 2026, all three hyperscalers shipped agentic-AI migration tooling. McKinsey estimates these tools compress migration timelines 30–40% (cited by AWS Transform team, Mar 2026). Most consulting firms haven't repriced. That's a negotiating lever.

AWS

AWS Transform

Agentic AI for VMware migrations went GA December 2025. Partners report 50% faster discovery, 80× faster network-config conversion. Replaces AWS Migration Hub, which closed to new customers November 7, 2025.

Source

Azure

Copilot Migration Agent

GA November 2025. Integrated with GitHub Copilot for code-level modernization (Java, .NET legacy stacks). Strongest in lift-and-modernize for VM-heavy estates already running Microsoft tooling.

Source

Google Cloud

Migration Center + Gemini

App Modernization Assessment GA September 2025. Gemini-powered codebase analysis flags refactor candidates in days, not weeks. Particularly strong for Java, Python, Go monoliths.

Source

Negotiating lever: Ask every prospective partner whether they're using AWS Transform, Copilot Migration Agent, or Migration Center as part of discovery. If yes, ask whether their pricing has been adjusted for the McKinsey-cited 30–40% time compression. Most firms still quote 2024 discovery rates against 2026 tooling. That gap is your discount.

Counterpoint

When not to migrate (or when to repatriate)

Vendor pages don't print this section. We will. Three signals tell you a workload should stay on-prem or come back from the cloud:

  • 1.
    Steady-state high utilization. AWS enterprise-strategy research cited by iternal.ai places the cloud-vs-on-prem break-even at roughly 20% sustained utilization over three years. AI training, batch ETL, and steady-state databases over that line are the workloads quietly going home.
  • 2.
    Sensitive AI workloads. The Cloudian 2026 survey reports 91% of regulated AI workloads end up on-prem or in private cloud. It's no longer a contrarian position — it's the median path.
  • 3.
    Predictable storage growth. Egress fees and per-GB storage scale linearly with growth. If your storage is doubling annually and the workload doesn't need cloud elasticity, you're paying a 30–60% premium for flexibility you don't use. Read Vendor Lock-in in Cloud Computing for the dollars.

Roughly 21% of cloud workloads are estimated to have been repatriated globally per Shopify's 2026 cloud-reset analysis; Gartner expects 90% of organizations to operate hybrid by 2027. The honest output of any readiness assessment includes "stay" as a valid result.

Hiring Guide

Match your readiness score to the right type of partner

Different score patterns need different shops. Hiring a global SI for a 50-app rehost is a budget grenade; hiring a boutique for a 7-month DORA migration is an ops grenade. Match the partner to the score.

Score sum 5–9

Assessment partner

You're early. Hire a discovery and assessment specialist before any execution SOW. 4–8 week engagement, $25K–$75K. They produce the dependency map, the risk register, and the strategy recommendation. They do not execute.

Look for: automated discovery tooling (AWS Transform, Azure Migrate, ADDM, CAST), published methodology, no MSP upsell pressure.

Score sum 10–17

Migration factory

Most mid-market shops live here. Pick a firm with a documented "factory" — repeatable wave templates, automated cutover patterns, blue/green tooling, post-migration MSP. 3–6 month execution, $200K–$500K.

Look for: 50+ comparable migrations, hyperscaler Premier/Expert tier, signed BAAs/DPAs ready to go on day 1, repatriation playbook (not just a migration playbook).

Score sum 18–25

Modernization specialist

Heavy lock-in tolerance, mature ops, and time pressure together usually mean you need refactor expertise on a tight clock. Boutique modernization or a hyperscaler-aligned specialist (AWS Premier modernization, Google Cloud refactor specialists). 6–12 months, $500K–$1.5M.

Look for: reference customers in your industry at comparable refactor scale, framework-specific experience (Spring, .NET Core, Django), opinionated architectural point of view.

For a deeper read on the partner-evaluation step, see our 10-point vendor due diligence checklist and guide to cloud migration consulting services.

2026 Directory

Cloud Consulting Firms Compared

Listed alphabetically — we don't rank firms by a hidden score. How we evaluate →

$200K+ typical · 500-1000 employees · AWS Premier Partner
AWSAzureGCP
View →
$250K-$1M+ · 10,000+ cloud specialists globally · Premier Partner
AWSAzureGCP
View →
$200K-$800K · 6,000+ AWS-certified · Premier Partner
AWSAzureGCP
View →
Caylent Featured
$200K+ typical · 200-500 employees · AWS Premier Partner
$50K-$500K (3PAO) · $750K-$2M (full Moderate ATO program) · ~1,000 employees, 100+ frameworks supported · Compliance & FedRAMP Specialist
AWSAzureGCP
View →
$180K-$600K · 5,000+ AWS professionals · Advanced Partner
AWSAzureGCP
View →
$300K-$2M+ · 8,000+ cloud professionals · Premier Partner
AWSAzureGCP
View →
$25K-$150K (retainer) · custom for active IR · enterprise-negotiated · ~1,400-2,000 consultants and analysts · Incident Response Leader
AWSAzureGCP
View →
$200K-$700K · 4,000+ cloud certified · Advanced Partner
AWSAzureGCP
View →
$100K-$400K · 180+ Google specialists · Advanced Partner
$100K-$500K (project) · MDR subscription OPEX · ~3,500-5,000 employees · AWS Premier Partner + MDR
AWSAzureGCP
View →
$150K-$550K · 400+ data specialists · Advanced Partner
GCPAWSAzure
View →
$150K+ typical · 5,000+ employees · AWS Premier Partner
AWSAzureGCP
View →
Slalom Featured
$250K+ typical · 10,000+ employees · AWS Premier Partner
AWSAzureGCP
View →
$180K-$650K · 3,000+ Azure specialists · Premier Partner
AzureAWS
View →
$150K-$500K · 12,000+ cloud specialists · Advanced Partner
AWSAzureGCP
View →
$160K-$550K · 8,000+ AWS certified · Advanced Partner
AWSAzureGCP
View →

Cloud migration consulting pricing benchmarks (2026)

Typical ranges from our partner data and Q1 2026 market analysis. Adjust the agentic-AI tooling discount (10–15%) before signing.

Engagement Type Price Range Typical Timeline
Discovery & Assessment $25K – $75K 4 – 8 weeks (1–2 weeks with agentic AI)
Mid-Market Migration (50–500 servers) $200K – $500K 3 – 6 months
Enterprise Migration (500+ workloads) $500K – $2M+ 6 – 18 months
Managed Services (post-migration) $5K – $50K/mo 12+ months (ongoing)
FinOps & Cost Optimization $10K – $30K/mo Ongoing or project-based

Hourly rates: $150–$200/hr (mid-market specialists) · $200–$300/hr (global SIs) · $100–$150/hr (offshore-led delivery). Rates reflect 2026 US market data.

Frequently asked questions

What are the 7 types of cloud migration (the 7Rs)?

The 7Rs are: Retire (decommission unused workloads), Repurchase (replace with SaaS), Retain (keep on-premises), Rehost (lift-and-shift to cloud VMs), Replatform (move with targeted optimizations like managed databases), Refactor (rebuild using cloud-native patterns), and Relocate (move between cloud regions or providers). They're not a menu of equal options — they're a decision sequence. A typical mid-market 100-app portfolio resolves to ~20% retire, ~15% repurchase, ~10% retain, ~30% rehost, ~20% replatform, ~5% refactor.

What are the 4 R's of cloud migration?

The original 4Rs (Gartner, 2011) were Rehost, Refactor, Revise, and Replace. AWS expanded these to the 6Rs in 2016 (adding Retain and Retire), and the 7Rs in 2021 (adding Relocate). The 4R framework is now considered legacy — most modern migrations need the full 7Rs to handle SaaS replacements, regulated retain decisions, and multi-region relocation. If a vendor is still selling you a 4R framework, that's a tell.

How much does cloud migration consulting cost in 2026?

Discovery and assessment: $25K–$75K over 4–8 weeks. Mid-market migrations (50–500 servers): $200K–$500K over 3–6 months. Enterprise migrations (500+ workloads): $500K–$2M+ over 6–18 months. Managed services post-migration: $5K–$50K/month. Hourly rates range from $150–$300/hour; global SIs charge at the high end, mid-market specialists charge competitively. Agentic-AI tooling (AWS Transform, Azure Copilot Migration Agent, Google Migration Center + Gemini) is compressing timelines 30–40% per McKinsey, but most firms haven't repriced yet.

How long does cloud migration take for a mid-market portfolio?

A mid-market migration (50–500 applications) typically runs 6–18 months end-to-end. Discovery and dependency mapping: 4–8 weeks (compressed to 1–2 weeks with agentic-AI discovery tools). Pilot wave: 2–3 months. Production waves: 4–12 months depending on dependencies. Post-migration optimization: ongoing. The single biggest timeline driver is application dependency mapping — Flexera 2026 lists it as the #1 challenge for the second year running.

Should I hire a single-cloud specialist or a multi-cloud firm?

Single-cloud specialists (AWS Premier Partners, Google Cloud Partners, Azure Expert MSPs) deliver deeper platform expertise, tighter access to hyperscaler engineering support, and faster ramp for target-platform deployments. Multi-cloud generalists (Accenture, Deloitte, Capgemini, IBM) are better for organizations running mixed workloads or those who haven't finalized a platform decision. If you've scored your readiness on our framework and the recommendation is Refactor or Rearchitect, lean toward a specialist. If it's Rehost or hybrid Replatform-Retain, a multi-cloud firm reduces vendor fragmentation.

What is the Migration Readiness Score and how does it differ from AWS MRA or Azure SMA?

Cloud Intel's Migration Readiness Score is a 5-axis vector — data sensitivity, vendor lock-in tolerance, ops maturity, regulatory load, and time-to-cutover budget — each scored 1 to 5. Unlike AWS MRA (~70 questions) or Azure Strategic Migration Assessment, our framework is read as a vector, not a sum. The combination of axes maps directly to a recommended primary R and secondary R, plus a 'delay' output when ops maturity is too low to migrate safely. It produces a strategy in five minutes rather than a maturity level after a 70-question questionnaire.

Should we be repatriating workloads from cloud to on-premises in 2026?

Possibly, for AI workloads. The Cloudian Enterprise AI Infrastructure Survey 2026 (n=203) found 93% of enterprises have repatriated AI workloads, are doing so, or are evaluating it. The break-even is roughly 20% sustained utilization over three years per AWS enterprise strategy research cited by iternal.ai. For traditional workloads — web apps, batch processing, microservices — repatriation is rarely worth the operational cost. The repatriation question matters most when you have steady-state high-utilization workloads and predictable storage growth.

What questions should I ask a cloud migration firm before hiring them?

Ask: (1) Show me a reference customer at comparable scale and complexity in my industry. (2) Walk me through your dependency mapping tooling — manual or automated? (3) What's your rollback plan if cutover fails? (4) Are you using agentic-AI migration tools (AWS Transform, Azure Copilot Migration Agent, Migration Center + Gemini)? If yes, has your pricing been adjusted to reflect the 30–40% time savings McKinsey reports? (5) What does your post-migration support model look like — SLA, on-call, drift detection? (6) Show me your repatriation playbook — when do you advise clients NOT to migrate? Firms that can answer all six with specifics and references are the ones worth shortlisting.