Data Cleanup Before ERP Migration: A 5-Step Plan
The most expensive misconception in ERP migrations: "The consultant will clean the data while migrating anyway." Data migrated without cleanup reproduces the old errors faster in the new system: risk limits split across duplicate customer records, inventory inflated by non-standard item codes, automations that fail on missing fields. In AI-assisted systems this becomes even more critical: a model cannot exceed the quality of the data that feeds it. In its ERP predictions report, Gartner projects that 70% of organizations will lack AI-ready ERP data by 2027. The good news: data cleanup is not a talent but a discipline — manageable with a five-step plan.
Why Before the Migration?
Cleanup during migration is the most expensive kind: done under schedule pressure, in the rush of field mapping, with "we'll fix it later" decisions — and "later" never comes. Cleanup done before migration is cheaper and speeds up the migration itself: what to migrate becomes clear, mapping gets simpler, test data becomes trustworthy.
The 5-Step Data Cleanup Plan
1. Inventory and ownership
Which master-data domains do you have (customers, products, suppliers, BOMs, price lists), and who owns each one? A data domain without an owner is a domain nobody cleans. First deliverable: a domain × owner × record-count table.
2. Define the standards
Coding rules (item-code structure, customer numbering), mandatory fields, format standards (phone, tax ID, units) are written down. Cleanup without a defined standard is cleanup by personal taste — the next person re-pollutes it.
3. Deduplicate and retire dead records
The same customer under three spellings on three records; item codes with no movement for five years; open records of departed staff. The rule: duplicates get merged, dead records get archived — not deleted (history reports need them), but not migrated as active either.
4. Enrich and validate
Missing mandatory fields (tax office, payment terms, unit weight) are completed; critical fields are validated at the source. This step's output is measured: a "mandatory-field completeness rate" tied to a pre-migration target.
5. Build the sustaining mechanism
Cleanup is not one-off: record-creation rights get narrowed, entry rules are configured in the system, periodic (monthly/quarterly) data-quality checks are assigned to the owners. Cleanup without a mechanism is back where it started within six months.
3 Common Mistakes
- The "consultant will clean it" assumption: the consultant maps and migrates; only your team knows which of your records is the correct one. Cleanup cannot be outsourced — only managed.
- Migrating everything: every dead record migrated "just in case" pollutes the new system's reports from day one. The archive stays in the old system or a data warehouse.
- Leaving cleanup to the end of the project: if test scenarios run on dirty data, the test itself becomes untrustworthy — cleanup must finish before testing.
Conclusion
Data cleanup is the highest-return line in an ERP budget: at relatively low cost it reduces both migration risk and the cost of errors in operation. To see where to start, the 2-minute ERP Risk Score shows your risk on the data dimension, and the free mini digital maturity scan your overall picture.
Related articles: An X-Ray of AI Promises · How Much Does an ERP Project Really Cost?
To catch new analyses, subscribe to the MindDX-Digital Excellence newsletter.