Data Migration

Short answer: Data migration happens when a company replaces or combines systems. Examples include moving to a new finance and operations system, sales system or job-management system, or bringing an acquired company onto shared systems. It is usually a project with a defined end. Data integration, by contrast, keeps data flowing between systems on an ongoing basis. Migration is also narrower than cloud migration, which moves applications and infrastructure as well as data.

A key early decision is scope: which records and how much history to bring. Many companies migrate their active core records, such as customers and products, plus open orders and a limited period of history. They archive the rest in a form that people can still access. There are two broad approaches. In a single cutover, often called big bang, everything moves at once over a short window. In a phased approach, data moves in stages while both systems run. A single cutover is simpler to manage but concentrates risk and usually needs downtime. A phased approach spreads risk but requires keeping two systems in step for a period.

The team first counts the records in the source system and checks their quality. It then maps each field in the old system to the new one, including splitting or combining fields and converting codes. It cleans records before transfer. It runs trial loads into a separate practice copy of the new system, so mistakes do not affect live work. It then compares record counts and totals between old and new. Business users test real processes on the migrated data. The final cutover follows a plan. The plan covers a freeze on changes, the final load and sign-off. It also covers a way to return to the old system if checks fail.

A field services company is moving to a new job-management system. The old system holds 6,000 customers, but only 4,200 have booked a job in the last 3 years. The team migrates those 4,200 and archives the other 1,800. The old system also holds 900 open jobs. After a trial load, the new system shows 897. The 3 missing jobs trace back to a customer code that failed mapping. The team fixes the code and reloads. The operations lead signs off only after the counts match, so no booked job goes missing on the first day.

The quality of migrated data decides whether a new system works on its first day. Missing customers, lost open jobs or broken product codes disrupt orders, scheduling and customer service. Migration is a common source of delay and cost overruns in system projects. It deserves its own plan, budget and business owner.

Migrations often take longer than planned because data problems appear late, when the team loads real records. Old history may not fit the new system structure. Legal and tax retention rules mean archived data must remain accessible for set periods. Many companies keep the old system available in read-only mode for a time. Migration is also a chance to stop keeping personal data the company no longer needs.