In short: on the French market, rebuilding a legacy application costs from around €10,000 for targeted refactoring to €75,000 and well beyond for a full rewrite, over a few weeks to two years. Start with a 3-to-5-day audit: it tells you which option fits. Most of the time, it's an incremental migration.
What is a legacy application, and when should you rebuild it?
A legacy application is software that is still useful, often critical, but has become hard to evolve. Its age matters less than its symptoms: nobody fully understands the code any more, dependencies no longer get security fixes, every change takes weeks and the team dreads each release. A typical case: an abandoned framework such as AngularJS, whose official support ended in January 2022.
Rebuilding doesn't mean throwing everything away. Depending on the state of the code, the work ranges from a simple hosting move to a full rewrite, and the budget spans the same gap. Before asking for a quote, it pays to know which option you're talking about.
What are the options for rebuilding a legacy application?
Four options come up in almost every project, and they can be combined. Replacement by an off-the-shelf product is a different decision, left aside here.
Rehosting: moving the hosting without touching the code
The application is moved as is to new infrastructure, for example from an in-house server to the cloud. AWS lists this approach (lift and shift) among its 7 migration strategies: the application changes servers without being modified. It's fast and low-risk, but the debt stays intact. You get better hosting for code that is just as hard to change.
Refactoring: cleaning up the existing code
The code is improved without changing what users see: putting critical areas under test, splitting tangled parts, updating dependencies. It's the cheapest option at each step, and it runs alongside day-to-day changes. But if the framework is no longer maintained, refactoring postpones the problem without solving it.
Incremental migration (strangler fig)
Incremental migration means replacing a legacy application piece by piece while it stays in production. Martin Fowler named it the strangler fig, after the fig that grows around a host tree until it replaces it. In practice, a layer is placed in front of the old system to route each request: features that have already been rewritten go to the new code, the rest are still served by the old one. You migrate one module at a time, starting with the one that causes the most trouble or brings the most value. Each step is covered by regression tests and can be rolled back. Your users never go through a big switchover, and you can slow down or redirect the work at any time. The trade-off: two systems living side by side during the transition, and glue code that gets thrown away at the end.
The full rewrite
You start from a blank page with a new stack, then switch every user over on a given date. On paper, it's the cleanest solution. In practice, it's the most expensive and the riskiest: the business rules buried in the code have to be rediscovered, and the old version has to be maintained until the switchover. It makes sense when the technology can no longer be run at all, or when the business has changed so much that the old scope no longer serves as a reference.
How much does a legacy application rebuild cost?
The cost of a rebuild is calculated in person-days, multiplied by the team's day rate. On the French market, a senior freelance developer generally charges €500 to €700 per day. The table gives orders of magnitude for a mid-sized business application, with a few dozen screens and two or three integrations.
| Option | Person-days | Benchmark budget (excl. VAT) | Timeline and risk |
|---|---|---|---|
| Rehosting | 5 – 20 | €2,500 – €14,000 | 1 – 4 weeks, low risk |
| Targeted refactoring | 20 – 60 | €10,000 – €42,000 | 1 – 3 months, low risk |
| Major framework upgrade | 30 – 120 | €15,000 – €84,000 | 2 – 6 months, medium risk |
| Full incremental migration | 80 – 250 | €40,000 – €175,000 | 6 – 18 months, medium risk, controlled in steps |
| Full rewrite | 150 – 400 and up | €75,000 – €280,000 and up | 9 – 24 months, high risk |
French market benchmarks for 2026, calculated with a senior day rate of €500 to €700, excluding the audit and hosting. They don't replace a quote: the audit exists precisely to cost your own application.
The main takeaway: a rewrite is rarely the economical option, because it reproduces years of business rules and forces you to run two systems until the switchover. For comparison, our article on the cost of an MVP puts a new product between €20,000 and €60,000 in most cases. Rewriting an application that has been in service for years generally costs several times that. An incremental migration can cost as much in total. It does, however, spread the spending, deliver value from the first months and can stop the day the gain no longer justifies the effort.
What makes the cost of a rebuild vary?
For the same size, the effort varies a lot from one application to another. Six factors weigh more than the number of screens.
- Without automated tests, every change has to be made safe first. Putting an area under test before touching it absorbs a good share of the initial budget.
- When the original developers have left and nothing is documented, the business rules have to be dug out of the code. It's slow, and it's where the nasty surprises come from.
- The data in a database with fifteen years of history, duplicates and changing formats needs migration scripts, checks and several rehearsals before the switchover. It's an easy line item to underestimate.
- Every integration (ERP, single sign-on, partner APIs, accounting exports) has to be rebuilt and then tested end to end.
- Tooling weighs in before the first line is even changed. An application spread across four or five repositories (website, mobile app, front end, back end, admin) already takes real work just to install and configure. Consolidating most of it into a monorepo is often the first step that pays off.
- An application used around the clock, or subject to healthcare or public-sector requirements, calls for maintenance windows, rollback plans and traceability that lengthen every step.
How long does a rebuild take?
Allow a few weeks for rehosting, one to three months for targeted refactoring, six to eighteen months for a full incremental migration, and nine months to two years for a rewrite. These timelines assume two or three senior developers, without freezing new features.
Look above all at when the first result arrives. An incremental migration puts its first module into production within a few weeks, whereas a rewrite shows nothing usable before the final switchover.
Why do big-bang rewrites go off the rails so often?
A big-bang rewrite replaces the whole system at once, on a set date. It goes wrong for two well-known reasons. The first lies in the code: the old application holds years of fixes and edge cases nobody documented. Starting from scratch loses them, and they get rediscovered one by one in production. The second is about timing: while the work goes on, the business keeps changing, and every new request has to be built twice or put on hold. Since value only arrives at the switchover, six months of delay means six months of cost with no benefit at all. Joel Spolsky described this trap back in 2000 with the example of Netscape, which went almost three years without a new major release during its rewrite (Things You Should Never Do, Part I). Rewriting is sometimes necessary, but it's rarely the right first choice.
What does doing nothing cost? The weight of KTLO
Doing nothing has a price too, called KTLO (Keep The Lights On): the time the team spends keeping the application running (fixes, incidents, forced updates, workarounds) instead of building new features.
In The Developer Coefficient, a study Stripe published in 2018 based on thousands of executives and developers in six countries, developers estimated that 17.3 hours a week are lost to maintenance in their company (bad code, debugging, refactoring), out of an average 41.1-hour week. France came out highest, at 20.9 hours.
To estimate yours, ask the team what share of its time went into maintenance last quarter, and apply that ratio to its annual cost. Put next to the rebuild budget, that figure speaks to management.
Why start with a 3-to-5-day audit?
An upfront audit costs little compared with a rebuild, and it saves you from costing blind. It is usually sold at a fixed price, over 3 to 5 days, which is roughly €1,500 to €3,500 at the market's senior day rate. The auditor reads the code, runs the application, reviews the infrastructure and talks with the people who maintain it. They map the debt: areas without tests, end-of-life dependencies, modules where incidents cluster. The deliverable is a prioritised action plan costed in person-days, which recommends an option (refactoring, incremental migration or rewrite) and the order of the work. You can then hand that plan to the team of your choice, including your own. Without an audit, a rebuild quote rests on assumptions nobody has checked. That is the format of our application technical audit.
Frequently asked questions about rebuild costs
Can you rebuild an application without stopping the business?
Yes, and that's the main benefit of an incremental migration. The application stays in production, each step can be rolled back, and sensitive cutovers are scheduled in slots agreed with you. A rewrite, on the other hand, forces a single switchover, with a rollback plan.
Should you change technology during the rebuild?
Not necessarily. Changing frameworks is expensive and only makes sense if the current one is no longer maintained or genuinely blocks your roadmap. Moving from Angular to React because React is more popular means paying for a rewrite for a theoretical gain. The stack your team already knows is usually the right one.
Fixed price or time and materials?
A fixed price suits well-bounded work, such as a version upgrade or migrating one specific module. An incremental migration over several months is better suited to time and materials (billing for time spent), because the scope becomes clearer as the code is uncovered. You can also fix the price of the audit, then of each migration batch.
Rebuild or refactor: do you have to decide up front?
No. That's the audit's job, and the two approaches are often combined. We go through the decision criteria in our article rebuild or refactor.
And at Neodev?
Our legacy application rebuild offer follows this method: a short audit, then a migration in reversible increments, with a system that stays in production at every step. We are used to working inside existing information systems. On the OPT NC customer portal (case study), for instance, part of the job was unifying authentication (OKTA) across all the applications.
After the audit, you receive a costed action plan and a detailed quote within 48 hours. The code remains entirely yours.
To go further: see how we help modernise a legacy application, or tell us about your application to get a first estimate.
Written by Adrien Le Breton, founder of Neodev. He has worked in software consulting and development since 2018, including more than two years in technical leadership.