How much it costs to modernize a 2000s custom ERP

Modernizing a custom ERP developed between 1998 and 2008 costs, in 2026, between 80,000 and 500,000 euros for an average Italian SME. The range is wide because the factors that move it (underlying technology, code size, external integrations, quality of historical data) work together and compound. Let’s look at how the estimate breaks down realistically, and why understanding it before asking for one prevents being sold something that doesn’t match the problem.

Answer in 3 lines

• Replatforming without rewrite (lift-and-shift on cloud, keeping the legacy stack): 60,000-150,000 euros in 4-8 months.
• Gradual refactoring (module-by-module rewrite with strangler pattern): 150,000-350,000 euros in 14-24 months.
• Replacement with a standard ERP (TeamSystem, Microsoft Dynamics, Odoo, SAP Business One): 200,000-500,000 euros in 18-30 months, licenses included.

The factors that determine the price

1. Underlying technology of the legacy

ERPs written in VB6, Delphi, classic ASP, advanced Access, PowerBuilder, or pre-AS/400 have different modernization costs. The most “dead” technologies (VB6, Access, FoxPro) cost more because finding developers who can decipher them is expensive: rates of 120-180 euros/hour against the 60-90 of modern stacks. Relatively newer stacks (.NET 2.0-3.5, Java 1.4-1.6, very early PHP) can be modernized more economically.

2. Code size and quality

Number of modules, function points, KLOC (thousands of lines of code). Above 100 KLOC, reverse-engineering times start to count significantly. Code quality is the hidden multiplier: well-structured and partially documented code can be modernized at 60% of the cost of “spaghetti” code with comments in dialect.

3. Number of external integrations

Each active integration (CRM, e-commerce, electronic invoicing via SDI, document management, warehouse, scales, barcode readers) adds 5,000-15,000 euros to the project. Integrations with Italian systems without modern APIs (old on-premise vertical business management systems, industrial scales with proprietary protocols) are the most expensive.

4. Volume and quality of historical data

Clean historical data migrates in a few weeks. Dirty historical data (duplicates, inconsistent encodings, empty fields with implicit meaning) requires data cleansing worth 10-25% of the total project cost. Above 50 million records, you also run into infrastructure problems (transfer times, indexing of the new DB).

5. Customizations accumulated for individual clients/sites

Custom ERPs from the 2000s often have 5-15 years of client requests integrated through localized workarounds (the Brescia site has its own version of the warehouse module, the sales office has its own custom dashboard, etc.). Each customization has to be decided: keep, discard, rewrite. On average, these decisions take up 15-25% of total project time.

6. Required operational continuity

If the system can stop for 1-2 weeks (e.g. extended weekend for go-live), costs are average. If 100% operativity is needed during the migration (strangler fig mandatory), costs go up by 25-40% because an intermediate compatibility layer has to be built.

Three concrete scenarios

Scenario A: custom ERP in VB6/SQL Server, 30 KLOC, 5 users

Manufacturing SME with an in-house ERP built in 2002-2005 in Visual Basic 6 with SQL Server 2005 backend. Around 30,000 lines of code, 4 modules (registry, orders, warehouse, invoicing), 5 active users, integration only with electronic invoicing.

Typical strategy: replacement with TeamSystem or Odoo, optionally preceded by 3 months of parallel running.

Cost: 90,000-160,000 euros in 8-12 months (of which standard ERP licenses 25-50k, implementation and data migration 50-90k, training and parallel running 15-20k).

Scenario B: custom ERP in Delphi/Firebird, 80 KLOC, 25 users, 6 integrations

B2B distributor with custom ERP in Delphi 7 developed between 1999 and 2010. Firebird/Interbase database. 25 active users, 6 integrations (B2B e-commerce, CRM, electronic invoicing, shipping management system, warehouse barcode readers, industrial scales).

Typical strategy: gradual refactoring with strangler fig, rewrite in Node.js/PostgreSQL on cloud architecture. The Delphi core is kept for 18-24 months while it gets replaced module by module.

Cost: 220,000-360,000 euros in 18-24 months (of which initial assessment 15-25k, new core development 130-200k, integrations 60-100k, data migration and training 30-50k).

Scenario C: custom ERP in .NET 2.0/3.5 + WPF, 150 KLOC, 80 users, 12 integrations, multi-site

Mechanical industry with custom .NET 2.0/3.5 ERP developed in 2007-2012, extended to a WPF edition in 2014. 150 KLOC total, 12 external integrations, 80 users across 3 Italian plants, specific job-order management processes.

Typical strategy: gradual refactoring + retirement of some modules + partial replacement with TeamSystem or SAP Business One for the standard part (accounting, HR, invoicing).

Cost: 380,000-550,000 euros in 24-30 months. This is the case where the “hybrid” strategy is almost mandatory: none of the pure ones works.

What’s included and what isn’t in a serious estimate

Things included in an honest estimate:

• Initial assessment (4-8 weeks of work): legacy documentation, dependency mapping, code and data quality evaluation
• Development of the new system (or configuration of the standard ERP if replacement)
• Historical data migration with documented cleansing plan
• Integrations with external systems agreed in the scope
• Functional and load tests
• Parallel running of 4-12 weeks (both systems active)
• End-user training (3-8 sessions depending on size)
• 3 months of heightened attention post-go-live (intensive bug fixing)

Things NOT included and to be quoted separately:

• Permanent software licenses of the new system (if replacement)
• Hardware or cloud infrastructure (variable based on choices)
• Scope extensions during the project (a change of requirements is another contract)
• Evolutive maintenance beyond the first 3 months
• Specific legal, fiscal, or compliance consulting
• Rewrite of business processes that were not in the legacy
• “Lifestyle” training (culture change, change management)

Red flags of wrong estimates

“We’ll rewrite it from scratch in 6 months for 80k”. Rewriting a functioning ERP from scratch in 6 months is a commitment that doesn’t hold up: real times are 12-24 months minimum for even just 80% of the original features. Whoever promises 6 months is either selling a prepackaged product passed off as custom, or giving an estimate that will explode mid-project.

“All-inclusive flat rate starting from X euros”. Custom ERP modernization can’t be flat-rated. Serious estimates are broken down by phase, with delivery milestones and economic check-points. The flat rate is almost always a way to dump risk onto the client.

“We guarantee zero loss of functionality”. Modernization always loses something, because some legacy features are accidental or no longer relevant to current business. A serious estimate starts with a mapping “what we bring / what we decide whether to bring / what we drop”, agreed with you.

“We’ll use AI to accelerate by 70%”. As seen in other blog articles, AI in modernization accelerates by 20-30% at most. Whoever promises 70% is selling a fantasy that will cost you for real when the project stalls halfway through.

“No assessment needed, we start right away”. Without an initial assessment of 4-8 weeks, any estimate is a guess from memory. The best modernization partners do assessments separately (5-25k euros standalone, discounted if the project proceeds). Whoever skips the assessment is either overestimating their knowledge of your case, or underestimating the complexity.

How a serious estimate is built in 4 steps

1. Non-binding discovery call (1 hour, free). What does your ERP do? How many people use it? Which integrations exist? In what language/framework? Without these basics no estimate is realistic.

2. Technical assessment (3-6 weeks, 5-25k euros standalone, discounted if we proceed). Reading the legacy code with modern tooling, dependency mapping, data quality evaluation, definition of the optimal strategy (replatforming vs refactoring vs replacement). Output: a 20-40 page document with cost/time estimate at confidence bands.

3. Detailed technical-economic proposal. Costs broken down by phase, delivery milestones, economic gates (e.g. end of phase 1 = 30%, end of phase 2 = 60%, go-live = 90%, end of bug fixing = 100%).

4. Optional POC on an isolated module (1-2 months, 15-40k euros). If the project is big (>250k), a POC on a non-critical module (e.g. client registry) validates the technical and team fit before committing to the whole.

FAQ

How much does data migration alone weigh in the total cost?

Typically 10-25% of the total, with peaks at 35% if historical data is very dirty or multilingual. A good estimate for your case is made after analyzing a sample of the DB: 1-2 weeks are enough to give a tighter range.

Is it worth waiting to hire internal staff to manage it?

Almost never. Internal staff is valuable AFTER modernization, for the evolutive maintenance of the new system. During the project, a specialized team is needed (senior developers with migration experience, data engineer, dedicated project manager) that is rarely sustainable as a permanent internal team. It’s worth doing modernization with an external partner + 1-2 internal people as technical link, and building the internal team afterwards for the run of the new system.

How much do you pay for parallel running?

Parallel running (both systems active in parallel for 2-12 weeks before cut-over) usually costs 15-25% of the total project cost: it includes double licenses, double data entry, reconciliation, doubled support. Expensive but almost always indispensable for critical businesses.

Is the new system always more expensive to run year over year?

It depends on the strategy. Replacement with a standard ERP has predictable recurring annual costs (licenses), typically growing 3-5% per year. Custom refactoring has variable evolutive maintenance costs (10-25% of the initial cost per year) but controllable. Replatforming has cloud costs that depend on volumes. Over a 5-year TCO the three approaches often converge, with different advantages (replacement = predictability, refactoring = ownership, replatforming = speed).

How long before ROI starts to show up?

For well-managed modernizations: 18-30 months from go-live, because ROI is mainly in: (1) reduction of staff dedicated to maintaining the legacy, (2) ability to adopt modern AI and automation, (3) reduction of operational risk. The first months post-go-live are often a valley of death where TCO temporarily rises before stabilizing.

Conclusion

Modernizing a 2000s custom ERP in 2026 is a serious project: we’re talking hundreds of thousands of euros, 1-3 years of work, impact on the entire organization. The expense is worth taking on if the legacy is becoming a bottleneck (talent, integrations, compliance, AI integration), not as a fashion. Once the decision is made, though, the quality of partner choice and strategy matters more than the budget number.

If you’re evaluating the modernization of your custom ERP and want a realistic estimate for your case, let’s talk. The first conversation is free.

For context: the pillar page legacy modernization, the page dedicated to custom ERP replacement, and the related article on AS/400 migration for those coming from that specific world.