Skip to content

Blog

Accounting Data Migration: Moving Financial Data Safely

How to migrate accounting and financial data between systems: what to move, opening balances vs full history, reconciliation, cutover timing, and the traps that break the books.

3 min read · Dino Correia
data-migrationfintechchecklist

TL;DR

  • Financial data has to balance after the move, to the penny, so reconciliation is the centre of the plan, not an afterthought.
  • Decide early between migrating full history or opening balances plus a read-only archive.
  • Cut over at a period end, and agree who signs off the numbers before you start.
  • Most problems come from mappings: tax codes, currencies, account structures, and records that only make sense in the old system.

Accounting data has one test that most data doesn’t: after the migration, the books still have to balance. Every account, every customer balance, and every tax total has to match, to the penny. That makes reconciliation the centre of the plan. The other big decision comes first: whether to move the full transaction history, or opening balances plus an archive of the old system.

The key decisions at a glance

DecisionOptionsWhat usually decides it
How much historyFull history, recent years only, or opening balancesReporting and audit needs, data quality, cost
Cutover dateAny day, or a period endPeriod ends make reconciliation far simpler
Chart of accountsKeep, map, or redesignWhether the old structure still fits the business
Open itemsMigrate individually or as balancesWhether you need to chase and match them later
Old systemSwitch off, or keep read-onlyAudit and retention requirements
Sign-offWho confirms the numbers are rightUsually finance, not engineering

Full history or opening balances?

Full history keeps everything in one place and lets you report across the move, but it means mapping years of transactions, including the odd ones, and reconciling all of them.

Opening balances move the balances at the cutover date, plus open items such as unpaid invoices and bills, and leave history in the old system or an archive. It’s much less work and risk. The cost is that reports spanning the move need both systems.

Many teams pick opening balances plus a read-only archive, then migrate a year or two of history only if reporting really needs it.

Mapping is where the books break

  • Chart of accounts. Account codes, types and hierarchies rarely line up one to one.
  • Tax codes. Rates, inclusive versus exclusive amounts, and country-specific rules.
  • Currencies. Which exchange rate applies to historical amounts, and how realised and unrealised gains are handled.
  • Contacts. Duplicates, merged records, and contacts that are both customers and suppliers.
  • Statuses. Draft, approved, part-paid, voided, written off: each needs a mapping, or a decision not to migrate it.
  • Records the old system created on its own. Rounding entries, system adjustments, automatic allocations.

Map every one explicitly, and decide what happens to records that don’t fit, before writing any migration code.

Reconciliation

Agree the checks with finance before you start:

  1. Trial balance at the cutover date matches between old and new.
  2. Customer and supplier balances match, per contact.
  3. Open invoices and bills match, individually.
  4. Bank balances match statements.
  5. Tax totals for the current period match.

Run them on every rehearsal, not just the final migration. More detail on layered checks is in data migration testing.

Timing the cutover

Cut over at a period end, ideally after the previous period is closed. Freeze changes in the old system for the cutover window, migrate, reconcile, and only then open the new system for new transactions. Plan for what happens to transactions that arrive during the freeze.

Keep the old system readable

Keep the old system, or an export of it, available read-only for as long as audit and retention rules require. You will need it, usually for a question nobody predicted.

The general version of this plan is the data migration checklist, and the choice between moving everything at once or in stages is covered in big bang vs phased migration. If you’re moving off a third-party platform, a migration and integration assessment maps the data and the risks before anything moves.

Where this leads

  1. If this is the situation you're in

    Migration & Integration Assessment →

    Understand data, dependencies and constraints before moving or connecting systems.

  2. How a problem like this was worked through

    Case study: Leaving a third-party platform →
  3. Or skip ahead

    Let's talk: tell me about the problem →

Related articles

Blog

Data Migration Checklist: What to Plan Before You Move

A practical data migration checklist: the questions to answer about source data, mapping, cutover, rollback, and validation before you move a single record.

· 6 min read
data-migrationsolution-architecturechecklist
Read →

Blog

Data Migration Testing: How to Prove the Data Is Right

How to test a data migration: reconciliation counts and totals, field-level sampling, business-rule checks, rehearsals against production copies, and sign-off criteria.

· 3 min read
data-migrationtestingchecklist
Read →

Blog

Big Bang vs Phased Migration: How to Choose

Big bang or phased migration? How each approach works, the risks and costs of each, the questions that decide it, and how to make a phased migration not drag on forever.

· 3 min read
data-migrationtechnical-decision-making
Read →

Working through something similar?

I help untangle problems like this one.