You don’t replace an ERP without first knowing what depends on it. We mapped every integration — inbound, outbound, scheduled, event-driven — and ranked them by risk before anyone bought a license.
Sanicare’s current ERP had grown fifteen years of organic integrations. Some documented, most tribal knowledge, all load-bearing.
Before any migration vendor walked in the room, the question was: what actually plugs into this thing, and what breaks if we move it?
Real behaviour drifted from any wiki page.
Built quickly, never decommissioned, still live.
No basis to scope a migration vendor against.
We combined static analysis (logs, schedulers, API gateways, network traces) with operator interviews — the people who actually keep things running.
The output was an integration register, a dependency graph, and a risk-ranked rebuild plan that the migration vendor could be scoped against.
Logs, schedulers, gateways. Behaviour, not docs.
Tribal knowledge → written register.
Rebuild cost, replace cost, risk if it breaks.