Nobody writes this one down, so here it is. This is what a Business Central go-live weekend looks like when it is done properly, hour by hour, for a typical mid-sized UK business going live on the first working day of a month. The detail changes with the business; the shape does not.
Two principles before the timetable. First, nothing happens on the weekend that has not been rehearsed. Every step below was run at least twice in the migration rehearsals, with timings recorded, and the weekend is the third performance. Second, there are two go/no-go points, and "no" is an acceptable answer at both. A rollback that costs a week is cheaper than a go-live that costs a quarter.
The week before
- Monday: go/no-go number one. UAT signed off, defects triaged (showstoppers fixed, the rest on the hypercare list), super users confirmed available, cut-over plan walked through line by line with named owners and expected timings against each step.
- Tuesday to Thursday: static data frozen in the old system (no new items, customers, suppliers, G/L accounts; anything urgent is logged for double entry). Users told, in writing, the last time they can post in the old system and the first time they can post in the new one. Printers, scanners, label printers, bank file formats and email sending tested from the production environment, not the sandbox.
- Friday daytime: normal trading. Last despatches confirmed. Finance posts everything they can. Warehouse does a stock count of anything doubtful.
Friday evening
- 17:00 – old system closed for posting. Someone with authority says so and the passwords are changed or the user accounts disabled, because "just one more invoice" is how every migration goes wrong.
- 17:30 – final extracts taken: trial balance, open customer and supplier items, stock by item and location (and lot or serial where tracked), open sales and purchase orders, fixed assets, bank balances. Control totals written down on one sheet. This sheet is the weekend.
- 18:00 to 21:00 – master data loaded into production: customers, suppliers, items, G/L, dimensions, posting groups, prices. Counts checked against the extracts. Any failures fixed and re-run, not worked around.
- 21:00 – stop. Tired people make data errors. The transactional load is Saturday morning.
Saturday
- 08:00 – opening balances loaded by journal: G/L balances to the suspense account, then open items to debtors and creditors, then stock by positive adjustment, then fixed assets, then bank. Each one posted and reconciled before the next starts.
- 11:00 – reconciliation one. Trial balance in Business Central equals the trial balance on the sheet. Aged debtors and aged creditors agree, in total and by customer or supplier. Stock value and quantity agree by location. Bank agrees. If anything is out, it is found and fixed now; nobody moves on with "we'll sort that Monday".
- 13:00 – open orders loaded: sales orders, purchase orders, production orders in progress, transfer orders. Spot checks against the old system by someone who knows the orders.
- 15:00 – integrations switched on, one at a time: web shop, EDI, bank feed, payroll journal, warehouse devices. Each one sends a real transaction through and it is checked at both ends.
- 17:00 – go/no-go number two. The project manager, the sponsor and the lead consultant look at the reconciliation sheet. Everything agrees, or everything that does not agree is understood, small and listed. If not: roll back. The old system is reopened on Monday, the lesson is learned, and the weekend is repeated in two weeks. It stings. It is the right call.
Sunday
- 09:00 – super users run the day-one scripts on production: raise a sales order, pick and ship it, invoice it, receive a purchase order, post a supplier invoice, run a payment, post a journal, print every document that will go to a customer. Real items, real customers, reversed afterwards or kept as the first transactions of the month.
- 12:00 – user accounts and permissions checked for every named user. Licences assigned. Everyone's role centre, saved views and report layouts are what they expect.
- 14:00 – the one-page "Monday morning" sheet goes out to every user: how to log in, what has changed in the three things you do most, who to call, and the first two days' support rota.
- 16:00 – everyone goes home. Monday starts early.
Monday
- 07:30 – consultants and super users on site, in the departments, not in a meeting room. Hypercare begins; we wrote it up separately in Hypercare: the first 30 days.
- 08:00 – first real orders. Someone watches the first ten of everything.
- 12:00 – stand-up: what broke, what is confusing, what is a defect, what is training. Fifteen minutes, every day for two weeks.
- 17:00 – bank posted, despatches confirmed, first day's transactions spot-checked. The sponsor is told, in one line, how it went.
When to call it off
You roll back at Saturday 17:00 if the trial balance does not agree and you cannot explain why; if stock does not agree by location; if an integration that the business cannot trade without is not passing transactions; or if a showstopper defect has appeared on production that did not exist in UAT. You do not roll back because of a long list of small things, a report layout that is wrong, or nerves. The list of small things is what hypercare is for.
And you never, ever go live with "we'll reconcile it next week". A trial balance that is out on day one is out forever, and every month-end afterwards starts with an argument about it.
If you would like your cut-over plan reviewed before the weekend, or someone senior in the room on the Saturday, get in touch.
Talk to usWant this done in your Business Central?
More from the blog
