Skip to content

How do you move your data to a new system without losing anything?

Short answer. List every place your data lives, decide how much history is worth bringing, then clean and de-duplicate customers, products and open jobs before anything moves. Map each old field to its new home, run trial migrations, and use the new system alongside the old one until the numbers match. Check counts, balances and open items, cut over outside month-end, keep the old system as a read-only archive, and name who signs off.

Two labelled moving boxes, one full of neatly packed folders and drives being carefully transferred into the other, red packing tape
In this guide · 5 minute read

Find every place the data lives

Start with an inventory, because the data that gets lost is the data nobody listed. The main system is obvious. The rest is not: the spreadsheet the service manager keeps for warranty jobs, the shared inbox where signed quotes arrive, the folder of PDFs on someone's desktop, the notes field in the old CRM that people use for everything.

For each source, write down what it holds, who maintains it, roughly how many records it has, and whether anything else depends on it. That list becomes the checklist you tick off at the end, so nothing is quietly left behind.

Childcare parent app with a photo story and the child's daily log
Parent app: the day's photos and daily log. Example screen; the name on it is invented.

Decide how much history to bring

Not everything needs to move. Open work always does: unpaid invoices, jobs in progress, live quotes, current stock, active customers and their contacts. Beyond that, bring what people genuinely look up. A service business may need every past job at an address, because technicians check what was done last time. A wholesaler may only need a couple of years of sales history for reorder levels.

Records you're required to keep, such as financial and employee records, must stay available for as long as the law requires, but they don't all have to live in the new system. A read-only archive of the old one can cover them, which keeps the migration smaller and the new system uncluttered.

Clean, de-duplicate and map before anything moves

Old systems collect duplicates: the same customer entered three times with slightly different names, products with two codes, jobs left open years after the work finished. Moving them across copies the mess into a system that was meant to be clean. Fix them at the source, with the people who know the records.

  • 01Customers: merge duplicates, agree which address and contact are current, and keep the old IDs as a reference.
  • 02Products and services: one code per item, retired items marked inactive, prices checked against the current list.
  • 03Open jobs and orders: close what's finished, confirm what's genuinely live, and match each to the right customer.
  • 04Field mapping: a written table showing where every old field lands in the new system, including the ones you've decided to drop.

★★★★★

“One of the BEST companies our business has worked with.”
Dhrumil V. · Tofulicious

Trial migrations, then run both systems

Run the migration as a rehearsal before it's real. Load a copy of the data into the new system, have the people who use it every day look through their own customers and jobs, and note what's wrong. Fix the mapping or the source data and run it again. Two or three rounds is normal, and each one is quicker than the last.

Then use the new system on real work alongside the old one. Quote, invoice and schedule in both for a period, and compare the results. When the numbers match, the history moves across for the final time and the old tool can be retired.

A real estate sales pipeline on a laptop in an agency
The agency's pipeline. Example screen; the name on it is invented.

The checks that prove nothing went missing

Reconciliation is how you know the migration worked, so agree the checks before the first trial run. Each one compares a figure from the old system with the same figure from the new one, and a difference is investigated until it's explained.

  • 01Counts: customers, products, jobs, invoices and attachments, in total and by type.
  • 02Balances: total owed by customers, total owed to suppliers, stock value, and a sample of individual account balances.
  • 03Open items: every unpaid invoice, live job and outstanding quote, listed in both and matched one to one.
  • 04Spot checks: twenty or thirty records chosen at random and compared field by field by someone who knows them.

Timing the cut-over and signing it off

Pick a quiet period for the final move. Avoid month-end, when invoices are being raised and accounts reconciled, and avoid the end of the financial year on 30 June. A date early in the month gives a clean break for reporting and time to sort out any problems before the next close. Keep the old system available read-only afterwards, so anyone can check an old record. Before the old system is retired, the person responsible for each area, usually finance, operations and the owner, signs off that the checks for their area match.

Terms in this guide

Questions owners ask us

How long should the old and new systems run in parallel?

Long enough for the numbers to match across at least one full cycle of your normal work. For many businesses that means a full billing cycle, so every kind of invoice, credit and payment has passed through both systems. Running in parallel takes extra effort, so set an end date and agree the checks that must pass before it. If the checks keep failing, extend it and fix the cause.

What if the numbers don't match after a trial migration?

Treat every difference as a clue, and find its cause before running the migration again. Most mismatches come from a handful of sources: duplicates that weren't merged, records filtered out by the mapping, rounding or tax rules that differ between systems, or transactions entered in the old system after the copy was taken. Fix the source data or the mapping, rerun the trial, and check the same figures again.

Who should sign off a data migration?

The people who own each area of the data should sign off, alongside the person running the migration. Finance should confirm balances and open invoices, operations should confirm jobs, stock and customers, and the owner or general manager should confirm the overall go-ahead. Each person signs off the checks for their own area, so if something later turns out to be missing, it's clear which check should have caught it.

Show us how you run it today.

Thirty minutes with the people who would build it. We'll map where the hours go and tell you honestly whether custom software is worth it for you.