Software Modernization Costs
Verified cost benchmarks for software modernization: median costs, cost ranges, timelines, and success rates across cloud, mainframe, data, application, and security modernization — aggregated from 3,733+ verified projects.
Software modernization projects range from targeted database migrations costing tens of thousands of dollars to multi-year mainframe-to-cloud transformations exceeding $15M. The benchmarks below aggregate cost, timeline, and success-rate data across 35 migration paths and 11 professional-services engagement types tracked by Software Modernization Intelligence, compiled from 3,733+ verified projects spanning cloud, mainframe, data, application, and security modernization.
All Migration Paths: Cost & Timeline Data
Every 35 tracked migration path with its verified cost range, median cost, median timeline, and success rate. Sort by any column, or filter to a single domain. Success rates are color-coded: green ≥80%, amber 65–79%, red below 65%.
How to read these benchmarks
- Cost range
- The low-to-high span observed across analyzed projects for that path. The spread is real, not a hedge — the gap between the floor and ceiling reflects genuine differences in codebase size, technical debt, and organizational scope. Expect to land inside the range, and treat the low end as achievable only when your project resembles the simplest projects in the sample.
- Median cost
- The central figure across the sample — a better anchor for a first-pass budget than either extreme of the range. Start here, then adjust for your own complexity, sector, and vendor tier.
- Median timeline
- Typical elapsed duration from kickoff to production cutover, not billed effort. Highly regulated environments and low test coverage extend this well beyond the median.
- Success rate
- The share of projects in the sample that met their primary objectives within acceptable cost and timeline overruns — not a guarantee, and not a claim that every aspect went to plan. Color-coding flags where a path is historically risky (below 65%) versus dependable (80% or above).
- Sample size (n=)
- The number of documented projects behind a path's figures. Higher counts mean more stable benchmarks; lower counts mean the range should be read as directional. Sample counts are shown per path and are not summed into a single headline, because one engagement can inform more than one path.
| Migration Path | Domain | Samples | ||||
|---|---|---|---|---|---|---|
| Azure → AWS | Cloud Modernization | $500k – $5M | $1.8M | 11 months | 78% | n=72 |
| EC2 (Lift & Shift) → Serverless (Lambda) | Cloud Modernization | $100k – $1.5M | $450k | 9 months | 80% | n=158 |
| GCP → AWS | Cloud Modernization | $400k – $4M | $1.4M | 10 months | 74% | n=58 |
| Heroku → Kubernetes (EKS/GKE/AKS) | Cloud Modernization | $20k – $500k | $100k | 4 months | 78% | n=50 |
| On-Premise → Hybrid Cloud | Cloud Modernization | $200k – $5M | $1.2M | 12 months | 75% | n=180 |
| VMware (vSphere) → Native AWS/Azure | Cloud Modernization | $500k – $10M | $2.5M | 12 months | 90% | n=124 |
| VMware vSphere → Nutanix AHV | Cloud Modernization | $500 – $2,500 per VM | $850 per VM | 5 Months | 65% | n=24 |
| Java EE (WebLogic) → Spring Boot (K8s) | Legacy Application Modernization | $500k – $5M | $1.2M | 18 months | 70% | n=142 |
| Monolith → Microservices | Legacy Application Modernization | $500k – $10M+ | $2.5M | 24 months | 40% | n=164 |
| Legacy PHP → Laravel | Legacy Application Modernization | $25k – $1M+ | $150k | 9 months | 85% | n=30 |
| Legacy Rails (v4/5/6) → Rails 7+ (Hotwire) | Legacy Application Modernization | $15k – $500k+ | $75k | 5 months | 92% | n=42 |
| AS/400 (IBM i) → Cloud | Mainframe Modernization | $500k – $4M | $1.5M | 16 months | 50% | n=95 |
| COBOL → C# (.NET) | Mainframe Modernization | $600k – $6.5M | $1.8M | 20 months | 68% | n=84 |
| COBOL → Java | Mainframe Modernization | $800k – $8.2M | $2.3M | 22 months | 62% | n=127 |
| Mainframe → AWS | Mainframe Modernization | $2M – $15M+ | $5.4M | 32 months | 55% | n=62 |
| Mainframe (z/OS) → Azure | Mainframe Modernization | $1.5M – $12M+ | $4.8M | 30 months | 58% | n=58 |
| Hadoop / AWS EMR / Snowflake / Legacy Data Warehouse → Databricks Lakehouse Platform | Data Modernization | $200K – $3.5M+ | $800K | 12-18 months | 60% | n=60 |
| MongoDB → PostgreSQL | Data Modernization | $150K – $2M+1 | $400K | 6 months | 55% | n=60 |
| Oracle DB → PostgreSQL (Vector) | Data Modernization | $25k – $1M+ | $150k | 6 months | 85% | n=179 |
| Legacy Data Warehouse → Snowflake Data Cloud | Data Modernization | $500K – $10M+2 | $1.5M | 18 months | 65% | n=220 |
| SQL Server → Python (Data Services) | Data Modernization | $50k – $800k | $250k | 8 months | 70% | n=172 |
| Data Warehouse (Teradata) → Snowflake / Databricks | Data Modernization | $200k – $10M+ | $1.2M | 18 months | 75% | n=150 |
| .NET Framework (Legacy) → .NET 6/7/8 (Core) | Legacy System Modernization | $50k – $1M+ | $180k | 5 months | 85% | n=112 |
| Delphi → C# | Legacy System Modernization | $200k – $2M | $600k | 15 months | 65% | n=70 |
| PowerBuilder → .NET | Legacy System Modernization | $300k – $2.5M | $850k | 18 months | 60% | n=85 |
| Visual Basic 6 → .NET 8 | Legacy System Modernization | $100k – $1.5M | $450k | 14 months | 72% | n=155 |
| Active Directory → Microsoft Entra ID | Security & Identity Modernization | $150k – $1.5M | $420k | 9 months | 82% | n=95 |
| Legacy SIEM (Splunk/QRadar/ArcSight) → Cloud SIEM (Sentinel/Chronicle) | Security & Identity Modernization | $80k – $2M | $350k | 9 months | 70% | n=42 |
| Legacy VPN (Cisco AnyConnect, Pulse Secure) → Zero Trust (Zscaler, Cloudflare, Entra Private Access) | Security & Identity Modernization | $40 – $200 per user/yr | $100/user/year | 9 months | 80% | n=120 |
| Redis → Valkey | DevOps & Platform Modernization | $0 – $500K+3 | $50K | 6 weeks | 60% | n=50 |
| SAP ECC → SAP S/4HANA | ERP Modernization | $300K – $400M+4 | $8M | 18 months | 40% | n=180 |
| AngularJS → React | Frontend Modernization | $50k – $500k | $150k | 6 months | 85% | n=168 |
| Oracle Forms → React | Frontend Modernization | $300k – $3M | $900k | 14 months | 60% | n=110 |
| React (SPA) → Next.js (SSR/SSG) | Frontend Modernization | $20k – $500k+ | $120k | 8 weeks | 92% | n=85 |
| Silverlight → Blazor | Frontend Modernization | $50k – $800k | $250k | 8 months | 90% | n=150 |
- 1MongoDB → PostgreSQL: Low end: small dataset (<100GB), simple schema, 1–2 months. High end: petabyte-scale data with complex nested documents, 12+ months.
- 2Legacy Data Warehouse → Snowflake Data Cloud: Low end: 500GB, 50 users, lift-and-shift. High end: 100TB+, 1K+ users, full re-architecture.
- 3Redis → Valkey: Low end: AWS ElastiCache in-place upgrade (Redis ≤7.2, no modules). High end: Redis 7.4+ with heavy Redis Stack usage at petabyte scale.
- 4SAP ECC → SAP S/4HANA: Low end: ~500 users, brownfield conversion. High end: Fortune 500 / globally complex greenfield re-implementation.
Professional-Services Pricing
Indicative pricing for all 11 modernization service engagements across three engagement sizes, plus the typical return-on-investment horizon. Actual pricing depends on scope, vendor selection, and organizational complexity.
| Service | Domain | Small | Medium | Enterprise | ROI | Typical Savings |
|---|---|---|---|---|---|---|
| Cloud Cost Optimization Services (FinOps) | Cloud Modernization | $30K - $60K | $80K - $150K | $200K - $400K+ | 3-6 months | 20-40% cloud spend reduction |
| Cloud Readiness Assessment & FinOps Strategy | Cloud Modernization | $80K - $150K | $180K - $350K | $400K - $800K | 6-12 months | $2M-15M annually |
| Kubernetes Migration Services | Cloud Modernization | $150K - $400K | $600K - $1.5M | $1M - $3M+ | 2-3 years | 20-40% infrastructure cost reduction via autoscaling and rightsizing |
| COBOL Migration Services | Mainframe Modernization | $80K - $150K | $180K - $350K | $400K - $800K | 18-24 months | $3M-15M over 5 years |
| Data Governance Strategy & Implementation | Data Modernization | $40K - $80K | $100K - $200K | $250K - $500K+ | 9-12 months | 20-40% reduction in data prep time |
| Legacy Application Assessment Services | Legacy System Modernization | $30K - $60K | $75K - $150K | $200K - $400K | 6-9 months | 15-25% Portfolio Rationalization |
| Zero Trust Architecture Design Services | Security & Identity Modernization | $40K - $80K | $100K - $200K | $250K - $500K | 12-18 months | 30-40% Network OpEx |
| Platform Engineering Services | DevOps & Platform Modernization | $80K - $150K | $200K - $400K | $500K - $1M+ | 9-12 months | 30% Developer Productivity Boost |
| Frontend Modernization Services | Frontend Modernization | $50K - $120K | $150K - $400K | $500K - $2M+ | 6-12 months | $200K-$1M annually |
| API Modernization Services | API & Integration Modernization | $40K - $80K | $100K - $300K | $200K - $500K+ | 12-24 months | 45% operational cost reduction, 70% performance improvement |
| Modernization Strategy Services | Modernization Strategy & Governance | $50K - $100K | $120K - $250K | $300K - $600K | 12-18 months | 20-30% IT OpEx |
Get the full modernization cost dataset
The complete benchmark workbook — every migration path and service, with cost ranges, median timelines, success rates, and sample sizes aggregated from 3,733+ verified projects. We’ll email it to you.
Cloud Modernization Costs
Cloud migrations span lift-and-shift datacenter exits, Kubernetes platform builds, and multi-cloud arbitrage moves. Cost variance is driven primarily by licensing landmines (Oracle and SQL Server per-vCPU cloud pricing) and data gravity — the time and cost to physically move large datasets to cloud storage. Rehosting is cheapest upfront but leaves unoptimized workloads that typically run 40% more expensive in the cloud than their refactored equivalents.
The number that decides your migration economics is your annual cloud bill, not your migration budget. Below roughly $500K/year of cloud spend, a rehost-then-optimize path almost always wins on total cost — the refactoring effort does not pay back before the underlying platform shifts again. Above $1M/year, the compounding waste of unoptimized workloads flips the calculation, and a refactor typically breaks even within 18–24 months. The most expensive mistake in this domain is treating a datacenter exit deadline as license to skip right-sizing: teams that rehost under time pressure inherit a cloud bill 30–40% higher than necessary, then pay a second engagement to claw it back. Budget for a FinOps discipline from day one rather than as a remediation project.
Observed cost span
Avg. success across 7 paths
Related Service Pricing (3 services)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| Cloud Cost Optimization Services (FinOps) | $30K - $60K | $80K - $150K | $200K - $400K+ | 3-6 months | n=420 |
| Cloud Readiness Assessment & FinOps Strategy | $80K - $150K | $180K - $350K | $400K - $800K | 6-12 months | n=340 |
| Kubernetes Migration Services | $150K - $400K | $600K - $1.5M | $1M - $3M+ | 2-3 years | n=380 |
- Cloud Cost Optimization Services (FinOps)— representative mid-size engagement
- Full FinOps Implementation. 8-12 weeks. Includes automated tooling (Finout/CloudZero setup), commitment modeling, and initial AI playbook.
- Cloud Readiness Assessment & FinOps Strategy— representative mid-size engagement
- 100-300 applications, multi-cloud (AWS + Azure), 3-5 data centers, complex compliance requirements. Typical for large enterprises ($2B-10B revenue).
- Kubernetes Migration Services— representative mid-size engagement
- Refactor to cloud-native (microservices architecture). 9-18 months. Includes full observability, GitOps, and FinOps implementation.
Full domain coverage: Cloud Modernizationhub →
Legacy Application Modernization Costs
Application modernization covers framework upgrades, monolith decomposition, and UI rewrites. Cost is driven by codebase size, test coverage (low coverage means expensive manual validation), and the complexity of embedded business logic that must be preserved exactly during migration. Two applications with identical line counts can differ threefold in cost based on their technical-debt profile.
The single biggest swing factor here is automated test coverage. A codebase with a comprehensive regression suite can be refactored aggressively because correctness is machine-verifiable at every step; a codebase without one forces manual validation of every changed path, which is slow, expensive, and error-prone. Before comparing vendor quotes, establish your coverage baseline — it explains more cost variance than line count. The second trap is the framework-upgrade that quietly becomes a rewrite: an AngularJS-to-React or Struts-to-Spring move often uncovers business logic fused into the presentation layer, and untangling it is the real work. Insist that discovery quantify how much logic lives outside a clean service boundary; that percentage, not the framework version gap, sets the price.
Observed cost span
Avg. success across 4 paths
Full domain coverage: Legacy Application Modernizationhub →
Mainframe Modernization Costs
Mainframe projects are the most expensive modernization category per line of code, driven by COBOL talent scarcity ($150K+/year for qualified developers), VSAM data complexity, and the mandatory requirement for bit-for-bit financial-calculation equivalence testing. Costs are measured per line of code or per MIPS consumed, not per application — you cannot lift individual components out of deeply interwoven batch sequences.
Equivalence testing is the line item that surprises first-time mainframe modernizers. In banking and insurance, the migrated system must produce byte-identical outputs to the legacy one across every code path before it can go live — not merely pass a functional test suite. Building the harness that proves this, sourcing representative production data, and reconciling the inevitable rounding and date-handling discrepancies routinely consume 25–40% of the total program. The second cost driver is the automated-versus-manual conversion decision. Tool-assisted COBOL-to-Java conversion is faster and cheaper per line but produces non-idiomatic code that is expensive to maintain; hand-rewriting produces clean code at several times the upfront cost. For a system you intend to keep for another decade, the maintainability of the output matters more than the conversion invoice.
Observed cost span
Avg. success across 5 paths
Related Service Pricing (1 service)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| COBOL Migration Services | $80K - $150K | $180K - $350K | $400K - $800K | 18-24 months | n=190 |
- COBOL Migration Services— representative mid-size engagement
- Multiple COBOL applications (50K-200K LOC), complex DB2 schemas, CICS transactions. Typical for regional banks or insurance carriers. 10-12 week engagement + optional 12-week pilot.
Full domain coverage: Mainframe Modernizationhub →
Data Modernization Costs
Database and data-platform migrations are frequently driven by Oracle and Teradata licensing costs — the recurring license burden averages 5–8% annual increases with $50K+ per processor core. Cloud data-warehouse migrations are well-trodden but routinely underestimate PL/SQL stored-procedure conversion complexity and BI dashboard re-validation scope. Stored-procedure count, not raw data volume, is the dominant driver of consulting hours.
The counterintuitive rule of data migration is that data volume is cheap and logic is expensive. Moving 50TB is a throughput-planning exercise measured in weeks of transfer over available bandwidth; converting 2,000 PL/SQL stored procedures with proprietary Oracle extensions is a labor-intensive translation exercise measured in consulting quarters. When you scope an Oracle-to-PostgreSQL or Teradata-to-Snowflake move, count the procedures, triggers, and packages first — that inventory predicts cost far better than terabytes. The chronically underbudgeted phase is BI re-validation: every downstream dashboard, report, and extract that depended on the old warehouse must be re-pointed and re-verified for numeric parity, and organizations routinely discover they own hundreds more of these than their catalog suggested. Licensing savings begin the day you decommission the source, which is why database migrations post the fastest ROI in this dataset — often 12–18 months.
Observed cost span
Avg. success across 6 paths
Related Service Pricing (1 service)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| Data Governance Strategy & Implementation | $40K - $80K | $100K - $200K | $250K - $500K+ | 9-12 months | n=310 |
- Data Governance Strategy & Implementation— representative mid-size engagement
- Strategy + Pilot Implementation. 12-16 weeks. Includes setting up a Data Catalog and training stewards for one domain.
Full domain coverage: Data Modernizationhub →
Legacy System Modernization Costs
Desktop application modernization (VB6, PowerBuilder, Delphi, WinForms) is quoted per form or per screen rather than per line of code, because COM/ActiveX dependency replacement and Crystal Reports migration are the primary hidden cost drivers — not raw code volume. A 50,000-line VB6 application can cost more to migrate than a 500,000-line Java monolith.
Desktop modernization defeats line-of-code estimation because the cost lives in the dependencies, not the application code. A VB6 or PowerBuilder application typically leans on a web of third-party ActiveX controls, COM components, and a reporting engine like Crystal Reports — each of which needs a modern replacement that behaves identically, and some of which are no longer sold or supported. The per-form pricing convention exists because a screen with a grid, three custom controls, and a printed report is a different unit of work than a simple data-entry form, regardless of how many lines back each one. When scoping, get the vendor to inventory external dependencies and reporting artifacts before agreeing a number; a quote based on form count alone, without that dependency map, is a guess that will be revised upward mid-project.
Observed cost span
Avg. success across 4 paths
Related Service Pricing (1 service)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| Legacy Application Assessment Services | $30K - $60K | $75K - $150K | $200K - $400K | 6-9 months | n=445 |
- Legacy Application Assessment Services— representative mid-size engagement
- Portfolio Assessment (20-50 apps). Automated scanning + high-level roadmap. 4-6 weeks.
Full domain coverage: Legacy System Modernizationhub →
Security & Identity Costs
Zero Trust and identity modernization costs are driven by identity-provider consolidation scope, the number of applications requiring Conditional Access policy authoring, and legacy protocol remediation (NTLM, Basic Auth). Microsoft Entra ID licensing ($6–$22/user/month) is the largest ongoing cost line item for organizations moving off on-premise Active Directory.
Identity modernization is priced per application integrated, not per user migrated, and the long tail of applications is where budgets break. Moving the workforce to a modern identity provider is straightforward; re-integrating every downstream application — especially legacy ones that only speak NTLM, Kerberos, or Basic Auth — is the slow, per-app work of authoring Conditional Access policies, testing access flows, and remediating protocols that predate modern federation. The ongoing cost is the licensing tier: advanced Conditional Access, identity protection, and privileged-access management sit in the higher Entra ID SKUs ($6–$22/user/month), so the per-seat figure your finance team sees is a permanent operating line, not a one-time project cost. Scope the application inventory and its protocol distribution first; the count of legacy-protocol apps predicts effort more reliably than headcount.
Observed cost span
Avg. success across 3 paths
Related Service Pricing (1 service)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| Zero Trust Architecture Design Services | $40K - $80K | $100K - $200K | $250K - $500K | 12-18 months | n=230 |
- Zero Trust Architecture Design Services— representative mid-size engagement
- Enterprise-wide Zero Trust Strategy. Includes Identity, Device, Network, and Data pillars. 8-10 weeks.
Full domain coverage: Security & Identityhub →
DevOps & Platform Costs
DevOps modernization costs are primarily professional services — CI/CD pipeline re-architecture, Internal Developer Platform (IDP) setup, and observability-stack deployment. There are few large migration-type projects in this domain; costs are scoped as ongoing platform-engineering engagements rather than one-time migrations, which makes them easier to phase but harder to bound.
Because there is no discrete artifact being migrated, DevOps and platform work resists fixed-price scoping and is best run as a capped time-and-materials engagement with clear milestones. The value — and the risk — is that the deliverable is a capability your team must operate, not a system that runs itself. An Internal Developer Platform that no one on staff can extend becomes shelfware within a year, so the decisive budget line is enablement: pairing, documentation, and a deliberate handover of platform ownership. Observability deserves its own scrutiny, because tooling license costs (log volume, metric cardinality, retention) scale with usage and can quietly become the largest recurring line item in the domain. Treat the platform engagement as buying a durable internal capability; if the plan does not fund your own team to run it, you are renting an outcome you will have to buy again.
Avg. success across 1 path
Related Service Pricing (1 service)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| Platform Engineering Services | $80K - $150K | $200K - $400K | $500K - $1M+ | 9-12 months | n=295 |
- Platform Engineering Services— representative mid-size engagement
- Full Platform Build. Multi-cloud support, 5+ templates, complex security integration. 3-4 months.
Full domain coverage: DevOps & Platformhub →
ERP Modernization Costs
ERP modernization is dominated by SAP ECC → S/4HANA, with Oracle E-Business Suite and legacy Oracle Forms making up most of the rest. Cost is driven overwhelmingly by the volume of custom code — bespoke ABAP or PL/SQL that must be remediated or rebuilt — and by the strategic choice between a brownfield technical conversion and a greenfield re-implementation. User count and the number of integrated satellite systems scale the figure further. This is the only domain on the site whose ceiling reaches nine figures.
The single decision that sets ERP cost is brownfield versus greenfield. A brownfield conversion lifts the existing configuration and data onto the new platform — the fast, lower-cost path (mid-size programs land in the low hundreds of thousands to low millions), but it carries forward every customization and all the accumulated technical debt. A greenfield re-implementation redesigns processes on a clean core; it forces the organization to confront two decades of customization, and for a global enterprise with thousands of users and heavy custom code the program can run to hundreds of millions and several years. The $300K–$400M+ span on the SAP S/4HANA path is not imprecision — it is the honest distance between a small brownfield conversion and a global greenfield transformation under the same product name. Inventory your custom code and settle your strategy before you read any vendor number, because that choice, not the license, moves the decimal point.
Observed cost span
Avg. success across 1 path
Full domain coverage: ERP Modernizationhub →
Frontend Modernization Costs
Frontend modernization covers framework migrations (AngularJS or jQuery to React, Silverlight to Blazor), rendering moves (React to Next.js), and rebuilds of forms-based UIs (Oracle Forms to React). Cost is driven by the number of distinct interactive components, the complexity of the application's state management, and how much shared design-system and API groundwork has to be built alongside the migration rather than assumed to already exist.
The trap in frontend work is treating it as a like-for-like port when it is really a rebuild of the interaction layer. A jQuery or AngularJS application usually has business logic and DOM manipulation tangled together, so the real effort is untangling state and re-expressing it in a component model — not translating syntax. Applications with a mature design system and a clean API boundary migrate quickly and cheaply; those without spend the first third of the budget building the component library and stabilizing the endpoints the new UI will consume. Count the distinct interactive components and the shared widgets before comparing quotes — a screen count alone understates the work whenever the current app lacks reusable primitives.
Observed cost span
Avg. success across 4 paths
Related Service Pricing (1 service)
| Service | Small | Medium | Enterprise | ROI | Samples |
|---|---|---|---|---|---|
| Frontend Modernization Services | $50K - $120K | $150K - $400K | $500K - $2M+ | 6-12 months | n=255 |
- Frontend Modernization Services— representative mid-size engagement
- Medium app (10K-50K LOC), multi-module frontend with complex state management. Typical for mid-market SaaS or e-commerce. 12-20 week engagement + optional rollout support.
Full domain coverage: Frontend Modernizationhub →
The Cost of Not Modernizing
Every budget on this page competes with the cost of doing nothing, which is rarely zero. The recurring premium of staying on legacy infrastructure is the other half of the business case — and it compounds for every year a decision is deferred.
Mainframe operations run 3–5× the equivalent cloud cost
Mainframe operations typically cost three to five times the equivalent cloud infrastructure per workload. That multiple is paid every year the migration is postponed, which is why mainframe programs carry the longest ROI horizon (3–5 years) yet deliver the largest long-term savings once complete.
Legacy database licensing escalates 5–8% a year
Oracle and Teradata license burdens average 5–8% annual increases at $50K+ per processor core. Because that escalating spend stops the day the legacy source is decommissioned, database migrations post the fastest payback in this dataset — often 12–18 months — and each year of delay is a year of avoidable license escalation.
Unoptimized workloads cost about 40% more to run
Workloads left unoptimized — whether rehosted under deadline pressure or never refactored — typically run about 40% more expensive than their optimized equivalents. Above roughly $1M/year of cloud spend, that compounding waste is what turns a refactor from optional into overdue.
What Drives Cost Variance
The ranges on this page span orders of magnitude within the same migration type. An Oracle-to-PostgreSQL migration can cost $25K or $1M. A cloud migration can cost $500K or $15M. Understanding why cost varies matters more than the headline number — because a vendor quoting the low end of a range without explaining their assumptions is the single most common cause of budget overruns. Seven factors account for most of the spread.
1. Codebase size and measurement unit
Backend systems are quoted per line of code. Desktop applications (VB6, PowerBuilder) are quoted per form or screen. Databases are quoted per schema, stored procedure, or terabyte of data. These units are not interchangeable — a 50,000-line VB6 application can cost more than a 500,000-line Java monolith because the VB6 forms carry deeply embedded business logic that cannot be converted automatically. Always establish the measurement unit before comparing quotes.
2. Technical-debt density
Two applications with the same line count can differ threefold in migration cost based on their debt profile. The signals that inflate cost: hardcoded configuration values (the team must find and externalize them by hand), undocumented external dependencies (discovery work before migration can even begin), absence of unit tests (validation becomes manual rather than automated regression), and circular service dependencies (components cannot be migrated independently). Vendors quoting without a discovery phase are guessing — and budgets built on guesses overrun.
3. Data complexity and volume
Data is often the most expensive part of a migration and the most consistently underquoted. An Oracle migration with 2,000 stored procedures and 50GB of data costs far more than one with 200 procedures and 500GB — because stored-procedure conversion is labor-intensive while data movement is largely automatable. Volume matters for throughput planning (a 50TB database takes four to six weeks to move over 1Gbps), but stored-procedure count, data-type mismatches, and referential-integrity complexity are what drive consulting hours.
4. Team readiness and skill gaps
A migration that lands in production where the internal team cannot operate the new system is not finished — it is deferred. Organizations that move to Kubernetes without building internal Kubernetes expertise typically call the vendor back within six months for incident support and optimization, effectively doubling the engagement cost. Budget for knowledge transfer (a minimum of 10% of total project cost), dedicated internal training (not optional shadowing), and a hypercare period where the vendor stays available after go-live.
5. Regulatory and compliance overhead
Regulated industries — financial services, healthcare, government — pay a 20–40% compliance premium on every modernization project. That covers audit-documentation requirements (every decision recorded and reviewable), change-control processes (every deployment gated by approval cycles that add two to four weeks per environment), and validation evidence (especially for banking mainframe migrations, where the system of record must produce identical outputs before and after). Regulated organizations should not benchmark themselves against the non-regulated median.
6. Vendor tier and commercials
Boutique specialists (10–100-person firms) typically charge 30–50% less per hour than enterprise systems integrators for the same technical work. The trade-off is capacity, contractual risk transfer, and procurement overhead. Enterprise SIs can absorb a failed migration; boutiques cannot. For projects under $500K, boutiques almost always deliver better value. For projects over $5M with regulatory exposure, the risk transfer and insurance value of an enterprise SI can justify the premium.
7. Contract structure and payment model
The commercial model moves the effective cost as much as the technical scope does. A fixed-price contract front-loads risk onto the vendor, who prices that risk in — a fixed bid on poorly-scoped work typically runs 20–40% above the equivalent time-and-materials engagement, and it creates an incentive to cut corners once the budget is consumed rather than absorb the overrun. Time-and-materials with a not-to-exceed cap aligns better on discovery-heavy migrations where scope is genuinely unknown up front, but it requires internal capacity to govern hours and sign off on scope. Milestone-based payments tied to validated outcomes — data reconciled, parallel run passed, cutover verified against the system of record — protect the buyer far more than calendar-based invoicing, because they withhold payment until the work is provably done rather than merely delivered. Finally, watch the tail: hypercare, knowledge transfer, and post-go-live optimization are routinely quoted as optional line items, then become mandatory the week after cutover. Budget them in from the start, in the original scope, rather than negotiating them as change orders from the weakest possible position once the vendor is the only party who understands the new system.
Before you request a vendor quote
- → Complete a discovery phase first. Vendors cannot give accurate estimates without one.
- → Specify your success metric. “Migration complete” means different things to different vendors.
- → Ask how they handle unknowns discovered mid-project. A fixed-price quote with no change-order process is a red flag.
- → Request references from migrations of similar size and complexity — not just logos.
Modernization Cost FAQs
How much does software modernization cost?
Costs vary enormously by scope and domain. A targeted database migration such as Oracle to PostgreSQL can run $25K–$1M. A full mainframe modernization program runs $2M–$15M+. The central tendency across all modernization types sits roughly in the $450K–$2.3M band, depending on codebase size, technical-debt density, and team readiness. Use the domain sections on this page for project-specific benchmarks rather than the aggregate.
What is the average timeline for a cloud migration?
Cloud migrations typically take 6–18 months. Simple rehost (lift-and-shift) migrations complete in 3–6 months. Replatform projects run 6–12 months. Full refactor or re-architect programs take 12–24+ months. Data gravity — the time needed to move large datasets — is the most common timeline surprise that pushes projects past their initial estimate.
Why do mainframe migrations cost so much more than other modernization types?
Three factors drive mainframe cost above all others. First, COBOL expertise is scarce — qualified developers command $150K+/year and booking lead times can exceed twelve months. Second, mainframe logic is embedded in batch sequences and VSAM file structures rather than discrete services, so individual components cannot be lifted out independently. Third, financial-calculation equivalence testing is mandatory — you must prove bit-for-bit identical outputs across every code path. A one-million-line COBOL codebase typically costs $1.5M–$4M and takes 18–36 months.
What are the biggest drivers of cost variance in modernization projects?
Five factors dominate. Codebase size — measured in lines of code for backend systems, forms or screens for desktop apps, and schema count for databases. Technical-debt density — undocumented dependencies, hardcoded configuration, and missing test coverage. Data complexity — stored-procedure count, data volume, and the rigor required for migration validation. Team readiness — whether your internal team can own the new architecture after go-live or needs extended vendor support. And vendor tier — boutique specialists cost less per hour but have capacity limits, while enterprise integrators charge a premium in exchange for risk transfer and contractual SLAs.
Is it cheaper to rehost (lift-and-shift) or refactor?
Rehosting is 30–50% cheaper upfront but generates higher ongoing cloud costs, because unoptimized workloads typically run about 40% more expensive in the cloud than their optimized equivalents. Refactoring carries a higher day-one cost but produces ongoing savings. The decision hinges on cloud-bill scale: below roughly $500K/year of cloud spend, the refactor ROI rarely justifies the difference; above $1M/year, refactoring typically breaks even within 18–24 months.
How should I budget for a modernization project?
Start from the median cost for your migration type, then add layers: 20% contingency for unknowns surfaced during discovery (dependency audits almost always reveal surprises), 10–15% for testing and validation (the most chronically underbudgeted line item), and 10% for change management and training. Always include a knowledge-transfer budget — if the vendor leaves and no one on your team understands the new system, you will spend the modernization cost again to recover.
What is a typical ROI timeframe for modernization investments?
Cloud migrations tend to break even in 18–24 months through infrastructure savings and licensing reductions. Database migrations off Oracle or Teradata onto open-source platforms break even faster — often 12–18 months — because licensing savings begin immediately. Mainframe modernizations carry the longest horizon (3–5 years) due to high upfront cost, but deliver the largest long-term savings, since mainframe operations typically cost three to five times the equivalent cloud infrastructure per workload.
How accurate is the cost data on this page?
All ranges are derived from analysis of 3,733+ verified projects. Median costs reflect actual documented outcomes where data was available, and ranges reflect real variance — the gap between a $25K and a $1M Oracle migration is genuine, driven primarily by stored-procedure complexity and data volume rather than vendor margin. Use this data for order-of-magnitude budgeting and vendor-conversation framing, not for fixed-price contracting.
How this data is compiled
Every range on this page is aggregated from 3,733+ verified projects — documented project outcomes drawn from published case studies, vendor disclosures, analyst reports, and procurement records, presented as ranges rather than point estimates. Actual costs vary significantly with organization size, codebase complexity, team capability, and vendor selection, so use these figures for order-of-magnitude budgeting and vendor-conversation framing — not for fixed-price contracting. Each per-domain table reports its own sample size (n=) so you can weigh recency and depth independently. Read the full Software Modernization Intelligence methodology for data sourcing, validation, limitations, and update cadence.
Read the methodology