Most migrations fail before the data moves. They fail at the export.

Start with what the law says you must keep

The ATO requires you to keep most business records for five years, in English or a form easily converted to it, and to be able to reconstruct your original data if your record-keeping system changes. It also expects digital records to be extractable into a standard format such as Excel or CSV, and notes that ASIC requires companies to keep records for seven years (ATO, record-keeping rules for business, checked September 2026).

Now compare that with what a conversion actually carries. Xero’s conversion partner, Jet Convert, brings across the current and previous financial year in its standard package — listed at $250 ex GST and currently subsidised by Xero to $0 — with each extra year of history $150, up to eleven years (Jet Convert AU pricing, checked September 2026, ex GST). Two years converted, five required. The gap is your archive, and it is your problem, not the new vendor’s.

The first item on the checklist is therefore not the new system. It is a complete export of the old one that you have opened, checked and stored somewhere the subscription cannot switch off.

Map the fields, then map the workarounds

Every business has a spreadsheet beside the system, doing the thing the system could not. The spreadsheet is the most honest documentation you own, and it will not migrate itself.

Write down every field in the old system and where it lands in the new one. Then write down every workaround — the spreadsheet, the naming convention in the job title, the colour code — and decide whether the new system absorbs it or you carry it across. Most businesses sit low on the ABS digital intensity index: 59% of those with up to four staff were in the lowest “baseline” band and 1% of all businesses reached “advanced” (ABS, Development of a Composite Indicator for Business Digital Intensity in Australia, 2021–22 data, article released 28 July 2023; the index is four years old, and we say so). Low intensity usually means more workarounds, not fewer.

This is also the moment to decide what not to move. Dead contacts, closed jobs older than the retention period, and half-finished quotes from 2022 cost the same to migrate as live data and pollute the new system from day one. Archive them in the export; do not import them.

Run one real job through both systems

Before the business moves, one job does. A real job — not a test job — from enquiry to paid invoice, entered in both systems for a fortnight. The real job brings its own mess, which is the point.

If a bridge is running between the two systems during the parallel period, understand how fast it moves. Zapier polls for new data every 15 minutes on its free plan, 2 minutes on Professional and 1 minute on Team (Zapier help, how Zap triggers work, checked September 2026). A bulk import into Xero runs against a limit of 60 API calls a minute and 5,000 a day per organisation, with 5 concurrent (Xero Developer, API limits, checked September 2026). Ten thousand contacts do not arrive in an afternoon, and a plan that assumes they do is a plan that finishes on Thursday with the office locked out.

ServiceM8 allows 180 API requests a minute and 20,000 a day per application and account, and returns HTTP 429 when you exceed either (ServiceM8 developer docs, HTTP response codes, checked September 2026). None of this is a reason not to migrate. It is a reason to schedule the import over days, not hours, and to know that going in.

Every integration to the old system breaks on cutover day

List every connection into and out of the old system before you switch: the accounting sync, the payment link, the form on the website, the automation someone set up in 2023 and forgot. Each one stops on cutover day, and each one needs rebuilding against the new system or deliberately retiring.

This is the part most migration plans miss entirely, because the person doing the migration is rarely the person who built the connections. Check what each system exposes before you commit — the integration checker covers the nine systems we have built against — and treat any connection to a system we have not built against as unknown until tested. Rebuilding the joins is most of what we do, and it is usually cheaper than the migration it follows.

• KNDR·01Export and verifyFull export, opened and counted. Attachments, notes, history and custom fields checked separately.
• KNDR·02Archive the gapEverything older than the conversion window, stored in CSV and PDF outside the old subscription, for the five years the ATO requires.
• KNDR·03Field map and workaround mapEvery field to its destination; every spreadsheet absorbed or deliberately kept.
• KNDR·04DateA dull Tuesday in a quiet month. Not BAS week, not end of financial year, not before a long weekend.
• KNDR·05Parallel jobOne real job in both systems for a fortnight, with the import throttled to the API limits.
• KNDR·06Training budgetHours × loaded hourly cost, per person, plus a slower fortnight.
• KNDR·07Integration listEvery connection to the old system, each one rebuilt or retired before cutover.
• KNDR·08Read-only periodThe old system kept readable until the first BAS on the new one reconciles.