Look at the records before you move them.
Moving into a new system is when you find out how inconsistent the old records have become. Some preparation turns those surprises into decisions made ahead of time.
What turns up when the files are compared.
One customer, three records.
“Acme Traders”, “Acme Trading Co.” and a phone number with no name attached, all the same buyer, each holding part of the order history. Someone has to decide which record survives and what gets merged into it.
Is 04/05 April or May?
Different people and tools save dates in different formats. Mixed formats in one column can quietly swap day and month, which matters for due dates, renewals and anything else with a deadline.
“Closed” meant several things.
Old systems pile up statuses over the years: closed, done, completed, cancelled-closed. Each one has to land on a status in the new system, and some won't have an obvious match.
Rows with nothing to join them.
Payments without an invoice number, orders without a customer code. With no shared identifier, records that belong together can't be linked automatically and need a rule or a person to decide.
From old files to a live system.
List every source
Spreadsheets, the old database, exports from other tools, shared folders, even a register someone keeps by hand. Note who looks after each one and how up to date it is.
Agree meaning and priority
When two sources disagree on a phone number or a balance, which wins? Settle what each field means and which source is trusted before anything is imported.
Match old fields to new
Write down how each old column and status becomes a field in the new system, including what happens to values that don't fit anywhere.
Run it on a copy
Import into a test environment and compare record counts, a sample of individual records, and the totals that matter, such as amounts outstanding or stock on hand.
Switch on a known day
Freeze the old system, run the final import, check it against the trial results, and start working in the new one.
Not every old record deserves cleaning. Spend the effort where the new workflow depends on it: active customers, open orders, current stock, outstanding balances. For historic or incomplete records, decide together whether they're imported as they are, archived somewhere searchable, or left behind with the original files kept safe.
Keep the originals untouched. Store a full copy of every source exactly as it stood before the move. When someone asks months later what a record used to say, you'll want to be able to check.
Settle the cut-off early.
When does the old system stop taking updates?
Pick the moment and tell everyone who enters data. Anything typed into the old system after it is either lost or has to be keyed in again by hand.
What if both systems run side by side?
Sometimes there's no way around it. Then the plan needs a way to capture changes made in the old system during that period, bring them across, and a named person to reconcile them.
Who signs off the result?
Ask the people who know the data best, often accounts or operations, to check the trial and the final import against figures they already trust.
What's the way back?
Agree what happens if the final import fails a check. Knowing you can fall back to the old system for a day takes much of the pressure off changeover day.
Moving off something old?
Application modernization
For when the old system itself needs replacing or rebuilding, along with its data.
See modernization Business needSoftware takeover
For when you're inheriting an application and its records from another team or vendor.
See software takeover GuideSoftware development costs
Why existing data is one of the harder parts of a project to estimate.
Read the guidePlanning a move to new software?
Send us a description of where your records live today and we'll talk through the migration.