Skip to main content
← All Insights
ERP Implementation6 min read

ERP Data Migration: Build Before the Data Is Yours

In a merger or carve-out, an ERP data migration waits on a legal date that keeps moving. A clean-sheet database lets the build start before data access.

An M&A ERP data migration is chained to a legal milestone. Until the deal closes, one side's systems stay closed to the other. That date has a habit of moving backwards, and every slip drags the integration plan with it. People who planned around the first date spend their weeks re-planning instead of building.

Most projects treat access to real data as the start line, so the migration clock only starts when the legal clock stops. Yet the work that fills the first weeks needs no real records.

This piece is about timing and preparation, not scope. ERP Data Migration: What Not to Move, and How to Sign Off covers the exclusion rules and the sign-off checks; ERP Data Migration: Master Data Is a Business Job covers who owns each data object. The three read as a set.

ERP Data Migration: What Not to Move, and How to Sign Off

ERP Data Migration: Master Data Is a Business Job

Legal close is not a project activity. It depends on regulators, signatures and lawyers, none of whom report to a steering committee. A date firm in spring is a target by summer.

It is not hypothetical. Mars announced its acquisition of Kellanova on 14 August 2024 and said at the time that it expected to close within the first half of 2025 (Mars to acquire Kellanova, 14 August 2024). The European Commission cleared it after a Phase II review on 8 December 2025, and the deal closed on 11 December 2025: more than five months past the company's own target (Mars and Kellanova, closed December 2025).

Mars to acquire Kellanova, 14 August 2024

Mars and Kellanova, closed December 2025

Each new date redraws the plan built on the last one. The date moves by six weeks. The cutover weekend is rebooked, the training week moves, and the key users who had freed up their calendars go back to their day jobs. The plan you presented to the board is now version two of a plan nobody trusts. The wider sequence is in The M&A Integration Checklist.

The M&A Integration Checklist

The real cost is sequencing. Hold the migration until access exists and your critical path runs: legal close, access, mapping, scripts, a first test load, then the dry runs. The only fixed point in that chain is the one nobody controls.

The clean-sheet database

Ask the incoming or outgoing ERP/CRM vendor to stand up a clean-sheet database: an empty instance with standard configuration and no live data, sitting harmlessly while the deal is negotiated.

It is a legitimate workaround, and it costs almost nothing. The instance normally sits inside the licence and build already being paid for. It holds no live records, so nobody involved sees data they are not entitled to.

What you build before the first real record

On a clean sheet, the first weeks are not idle time. They are when the thinking gets done.

  • Agree the data model. Which objects exist, which fields matter, which relationships the target expects. A model argued out before a single record arrives is one you can defend.
  • Explain the business meaning. Each object carries a meaning at the source that the target may not share. Explaining what a customer, an article or a price agreement represents to the people who will run the system prevents a technically perfect load of the wrong thing.
  • Design the mapping tables. Source field to target field, transformations written down, exceptions named. Most of the real analysis lives here, and needs no live data to start.
  • Write and debug the scripts against dummy data. A script that runs on invented records is ready for real ones, and a mistake is cheaper to find in week two than in the week the deadline lands.

By the time access arrives, the mapping is not being invented under pressure. It is being filled.

When access arrives

Share and explain the master data properly, rather than dropping an extract on a consultant. The people who know the records should say what is in them, what is missing and what has been stored in the wrong field. Then map it onto the existing tables, and correct what the dummy data could not show.

That is what makes a first test migration possible within 48 hours of getting access, instead of four weeks later.

The Belgian Waterlogic and Culligan merger

Ben Van Aken used this route on the Belgian Waterlogic and Culligan merger, working with Peggy Frans at Culligan Belgium, whose help was a large part of why the two-week migration landed. The mapping and the scripts were built on a clean sheet before access to the real data arrived.

Peggy Frans

The migration itself was finished in two weeks. The four weeks after were dry runs against real source data, and that is where the old contracts, the wrong entries and the records needing a decision all surfaced. The legal merger completed on 2 November 2022 and the system had to be live on 1 January: six weeks of work, and it finished inside that deadline. The mapping and the scripts were already built, so the first test load ran within 48 hours of access. The surrounding integration work is in the Culligan Belgium ERP and CRM merger case study.

Culligan Belgium ERP and CRM merger case study

What it buys, and what it does not

The clean sheet gets the plumbing built, not the data cleaned. Real-data validation still decides quality, and that still costs weeks, because filtering real records is where the old contracts and wrong entries appear and where somebody has to rule on each. What changes is when those weeks run: in parallel with legal, not after it.

The first three moves matter more than they look. Get the empty instance in the week the deal is announced, as a deliverable with a date. Put two people on mapping, one who knows the data and one who can write and test the scripts. And treat the legal date as a variable rather than a start date.

A short checklist

  1. Ask for a clean-sheet database as soon as the deal exists, not after close.
  2. Agree the data model and the business meaning of each object first.
  3. Design the mapping tables, and write them down.
  4. Write and debug the scripts on dummy data.
  5. Explain the master data when access arrives, then map it.
  6. Run the first test migration within 48 hours of getting access.
  7. Use the dry runs for filtering, validating and cleaning records.
  8. Keep the legal date out of the critical path.

The bottom line

Waiting for legal access before starting an ERP data migration hands your schedule to people who do not work for you. A clean-sheet database costs little, sits inside the licence and build already being paid for, and lets the model, the meaning, the mapping and the scripts be finished before the first record arrives. Build first. The data catches up.

Comments

Loading comments…

Facing a merger, a carve-out or a migration with a legal date on it?

Talk to Ben Van Aken, freelance ERP/CRM project lead, on-site, hybrid or remote. Details at swiftly-business-consulting.com.

Currently accepting new engagements, starting now and into 2027.