Utility Network migrations fail on data, not software. This playbook lays out the sequence that works: model mapping, data repair, asset packages, validation gates and a cutover that operations can live with.
Why migrations stall
Esri’s Utility Network rewards utilities with tracing, subnetwork management and integrity rules the geometric network never enforced — and that is exactly why migrations stall. Decades of connectivity shortcuts, missing phasing and orphaned features that the old model tolerated become hard errors in the new one.
Treating migration as an IT upgrade produces the classic failure: months of licence spend while the pilot domain still will not validate. Treating it as a data program with a software milestone at the end is what works.
Phase one: model mapping and honest assessment
Start by mapping your existing schema to the target asset package — every feature class, every attribute domain, every implicit rule your operators carry in their heads. The gaps this exposes are the real project scope.
Then measure the data honestly: connectivity break counts, phasing completeness, attribute fill rates on migration-critical fields. A defensible baseline turns the repair phase from an open-ended slog into a burn-down chart executives can follow.
Phase two: repair at production scale
Data repair is high-volume, rule-driven work — ideal for a dedicated production team running documented fix protocols with daily throughput reporting. Connectivity and phasing repair, device attribution, terminal configuration and containment relationships each become a tracked workstream.
The discipline that matters: every fix batch validates against the target rule set before it counts as done. Repairing to the old model’s tolerance just relocates the failure to cutover week.
Phase three: validation gates and cutover
Migrate domain by domain through explicit gates: asset package loads clean, topology errors below threshold, subnetworks trace correctly against known-good cases, and operations sign off on a parallel-run comparison. Only then does a domain cut over.
GlobeIR runs migration as a factory — model mapping, repair workstreams, asset package preparation and validation reporting — so your GIS team steers the program instead of drowning in it. Utilities we have migrated kept their cutover dates because the data was ready before the software was scheduled.