Most ERP implementations fail slowly rather than dramatically. They stall because the data was messier than expected, or because the person who could decide was never available, or because everyone went live on everything at once and no single module ever worked well enough to be trusted. A twelve-week plan with named owners and a narrow first scope avoids all three.
Weeks 1–2: scope and freeze
- Name one internal owner with decision authority.
- List the modules going live in phase one. For most developers: inventory and booking, kisti and collection, receipts. Nothing else.
- Freeze the process changes you intend to make. Automating a broken discount ladder just makes it fast.
- Agree the go-live date and the parallel-run month now, so they exist in people's calendars.
Deliverable: a one-page scope with a date and an owner.
Weeks 3–4: data cleanup at source
- Choose the authoritative source for each data set, and record the choice.
- Clean in the spreadsheets, where your team knows the data.
- Resolve duplicate buyers, units that appear in one list and not another, and bookings whose payment schedule was edited but whose agreement was not.
- Produce the baseline reconciliation: units, bookings, contract value, received, outstanding, per project.
This is the phase that determines the timeline. See migrating from Excel for the detail.
Weeks 5–8: configuration
- Projects, buildings, units, and the price rules that compute unit prices.
- Payment plan templates — expect four or five genuine ones.
- The discount approval ladder and who sits on each rung.
- Receipt types, payment channels, and the allocation rules for partial and excess payments.
- User roles, and specifically who may release a hold, approve a discount, and edit a schedule.
- The three reports management will actually open: due versus collected, ageing, and unit inventory status.
Configuration is done with the internal owner in the room. Configuration done by email takes twice as long and ends up wrong.
Weeks 9–10: load and reconcile
- Load master data, then transactions, then opening balances.
- Produce the reconciliation report against the week-4 baseline.
- Explain every difference. Not most — every one.
- Train the people who will use it daily, on their own data rather than a demo dataset.
Week 11–12: parallel run and go-live
- Run both systems for one full month including a month-end close.
- Compare collection for the month, ageing by bucket, and ten randomly chosen buyer ledgers.
- When they agree, cut over, and make the old files read-only from that day.
- Schedule the second training round for six weeks after go-live, when people have real questions.
What to protect
Two things fail more often than the software:
Time. Two to four hours a week from finance, sales and operations. Put it in calendars.
Decisions. Every implementation surfaces twenty questions of the form "which of these two is correct?". If those queue for a weekly meeting, the project takes six months instead of three.
What to do next
Before signing anything, produce the week-4 reconciliation for one project from your current spreadsheets. However long that takes is your real timeline, and whatever will not reconcile is the actual work — walk the first two phases on your own data.