D365 data migration decides most Finance and Operations go-lives, and the outcome is largely set before UAT starts. UAT finds configuration defects, which are fixable in days. It cannot fix a customer master full of duplicates, opening balances that fail to reconcile, or an extract-transform-load process that takes ninety hours when the cutover window is forty-eight.
Ask people what went wrong on a difficult F&O programme and you will hear about scope, the vendor, or change fatigue. The project artefacts usually show something more specific: data migration started late, ran as a side workstream, and was rehearsed properly for the first time a few weeks before cutover.
On our Dynamics 365 programmes, data migration is on the plan from the first week, staffed as an engineering workstream with its own people.
The ETL routines get built early and run often. Every run is measured on two numbers, error rate and runtime, so progress shows as a trend line. If the error rate is falling and the runtime fits the cutover window, the programme is on track in the way that matters most.
Master data comes first, because everything else hangs off it. Customers, vendors, items and the chart of accounts get extracted, cleaned and loaded before anyone worries about history. This is also the one structured opportunity the business will get to clean its data.
A cutover rehearsal is a full run of the real thing: extract from the live system, transform, load, reconcile, and time every step. The first rehearsal is always rough, and its purpose is to expose the problems. The value is in the improvement between rehearsals, which is why each one is measured against the last.
By the third or fourth run, the team stops debating whether cutover will work and starts arguing about whether the window is forty hours or forty-four, based on measured runs.
The final cutover should be the best-rehearsed activity in the entire programme, and on well-run programmes it is also the least dramatic.
The other decision that shapes migration effort is history. Open balances, open orders and inventory positions must move and must reconcile to the ledger. Years of closed transactions mostly should not, because loading a decade of history into an operational system adds effort and running cost for very little operational return. Reporting needs are better served from a warehouse or lakehouse, as covered in where historical ERP data should live. The rest of the migration path is in our AX 2012 to D365 migration guide.
If you are sponsoring an F&O implementation now, one question will tell you more than any status report: how many full migration rehearsals have run, and what were the error rate and runtime on the last one compared to the one before?
A programme that answers with numbers is in control. A programme that answers with a plan to start rehearsing soon does not yet have a reliable go-live date.
In the first week of the programme, with master data first and ETL routines running early enough to build a trend.
Plan for several full runs. By the third or fourth, the discussion is usually about a few hours of difference in the cutover window.
UAT tests configuration and processes. Duplicate master data, balances that do not reconcile and loads that overrun the cutover window only show up in full rehearsals with real data.
Marcel Rizzolo is Director of Coder Trove.