Legacy System Migration

Legacy system migration: a process you can actually plan against

Legacy system migration is the narrower discipline inside application modernization - moving from a specific legacy system to a specific successor without losing functionality. This page covers the migration programme as a BPMN process map.

Jack Finnegan, Founder & CEO, BA Copilot

By Jack Finnegan ยท Updated 21 May 2026

What it is

What legacy system migration actually is

Legacy system migration is the structured programme to move from a specific legacy system to a specific successor - data, integrations, business processes, users - without losing functionality or operational continuity. It is the narrower discipline inside application modernization (which also covers retain/retire decisions and rebuilds); migration is specifically the case where the answer is 'move to a new system, keep the functionality'.
Common migration shapes: on-premises ERP to cloud SaaS (Oracle EBS to Oracle Fusion Cloud, SAP ECC to SAP S/4HANA), legacy CRM to Salesforce, legacy data warehouse to Snowflake or Databricks, mainframe to distributed systems, end-of-life vendor to current vendor. The shape varies by domain; the programme structure does not.
The problem today

Migrations slip because nobody documented what the legacy actually does

The recurring story: a migration programme kicks off, the consulting firm runs requirements workshops, the build team starts on the target system, and three months before cutover the team discovers that the legacy handles a dozen edge cases that nobody documented. The cutover slips; the budget overruns; the workarounds that the legacy was quietly handling have to be added to the target as exception logic.
The fix is rigorous current-state process mapping before the design starts. BPMN diagrams of every in-scope business process - including the variants and exceptions the formal documentation hides - become the requirements specification for the target. Without that artefact, requirements emerge during the build, which is the most expensive moment to discover them.
Four pillars

Four pillars of a working migration

Honest legacy assessment

Functionality, data, integrations, technical debt, contractual constraints, end-of-life dates. The assessment shapes everything that follows.

Target scope discipline

Decide what functionality is in scope for migration vs what is being retired. Scope creep ("we always meant to add that") kills migration programmes.

Wave-based data migration

Migrate data in waves with reconciliation at each wave, not a single big-bang load. Big-bang loads are how you discover data quality issues at the worst possible time.

Rehearsed cutover with rollback

Rehearse the cutover in pre-production, including the rollback plan. The cutover is the highest-risk moment; treat it like a major release.

Process Map

A migration programme as a process map

The canonical flow - assess, target, design, build, migrate, cut over, decommission.

Open in editor

A legacy system migration programme as a process map

A canonical legacy system migration programme rendered as a BPMN 2.0 process. Assess the legacy, define target, design migration, build, migrate data in waves, cut over with rollback, and decommission.

  1. Assess the legacy system - functionality, data, integrations, technical debt, contractual obligations.
  2. Define the target system or platform and the in-scope functional set.
  3. Design the migration approach - big bang vs phased, data-mapping rules, integration cutover plan.
  4. Build or configure the target system to the agreed scope.
  5. Migrate data in waves with reconciliation at each wave.
  6. Cut over with a rehearsed rollback plan; run parallel for the agreed quiet period.
  7. Decommission the legacy system - access revoked, data archived, contracts closed.
What this diagram shows: The programme starts when migration is triggered - usually by a legacy end-of-life, a strategic technology shift, or a regulatory change. Legacy assessment shapes the migration design; target definition fixes the scope. Build precedes the wave-based data migration. Cutover runs with a rehearsed rollback plan. Decommissioning closes out the legacy - access revoked, data archived, contracts closed.
FAQ

Frequently asked questions

What is legacy system migration?

The structured programme to move from a specific legacy system to a specific successor without losing functionality. It is one specific shape inside the broader application modernization discipline.

What's the difference between migration and modernization?

Modernization is the broader discipline that includes migration plus retire, refactor, rebuild, and rehost decisions. Migration is specifically the case where the answer is "move to a new system, keep the functionality".

When should you do a big-bang migration vs phased?

Big-bang (cut everyone over at once) is faster and operationally simpler but riskier - all the risk concentrates on one cutover weekend. Phased (migrate by region, business unit, or capability) is slower and more expensive but spreads the risk and gives time to learn from each wave.

How do you handle data migration?

Best practice is wave-based: migrate a manageable slice, reconcile fully, fix any issues, then migrate the next slice. Big-bang data loads are how migrations discover their data quality issues at the worst possible time. Reconciliation is the single highest-leverage activity in a migration programme.

How does process mapping fit into migration?

Process mapping is how you produce the requirements specification for the target. BPMN diagrams of every in-scope business process - including variants and exceptions - drive the configuration or build of the target system. Without that artefact, requirements emerge during the build, which is the most expensive moment to discover them.

Does BA Copilot help with legacy migration?

BA Copilot accelerates the discovery half of the migration. Legacy assessment usually requires mapping dozens of business processes that the legacy system supports - work that historically takes weeks of analyst time. AI-assisted modelling collapses that to days, freeing the analysts to focus on the variants and exceptions where the value is.

Jack Finnegan, Founder & CEO, BA Copilot
From the founder

14 Years in BPMN

I'm Jack Finnegan. I've spent fourteen years working hands-on with BPMN, as an analyst, an engineer, and a product director, where I felt every sharp edge of legacy business process platforms.

BA Copilot is the platform I wanted on every one of these projects: AI-first process management, which treats BPMN as a first-class output rather than an export afterthought.

Cosmic background pattern
Decorative rectangle pattern

Make the legacy specification visible

Open BA Copilot, capture the legacy processes as BPMN with AI-assisted modelling, and produce the requirements specification the migration team will build against.