The Implementation, in Order

Implementation effort in this category is concentrated in configuration and reconciliation, not in installation. Plans that assume otherwise run late in a predictable place.

The sequence

One: configure the rules, with payroll present. This is the longest phase and it is usually planned as the shortest.

Two: load reference data — people, projects, cost centres — from the authoritative systems, by the integration rather than by hand.

Three: build and test the outbound integration, with real data shapes.

Four: migrate history, or decide to keep the old system readable instead.

Five: pilot with one group, through a full period including close.

Six: parallel run, at least three periods.

Seven: cut over, with the old system readable.

Eight: stabilise, which is a phase and not an afterthought.

Where the time actually goes

Rule configuration. Every exception, every grandfathered agreement, every "except for the night shift in the second site".

Reconciliation of the parallel runs, line by line.

Chasing reference data quality, which is always worse than anyone believed.

Training, which is small per person and large in total.

Not installation, which for a hosted product is an afternoon and which plans treat as the milestone.

The rules discovery problem

Most organisations cannot produce a complete list of their own pay rules.

They exist in payroll configuration, in collective agreements, in individual contracts, and in one person's memory.

Budget for discovery as a phase, with payroll and HR, before configuration.

Expect to find rules that contradict each other, and rules nobody can justify. Resolving those is an organisational decision, not a configuration task, and it is on the critical path.

What must be true before go-live

Three parallel periods reconciled line by line, signed off by payroll in writing.

The statutory record demonstrably producible for a named person and period.

Worker self-access working.

The export tested end to end, including a correction after export.

A rollback plan that someone has read.

Support arrangements live, not starting the week after.

The phase nobody plans

Stabilisation: four to eight weeks after cutover where corrections are frequent, questions are constant and someone must be available.

Plan capacity for it. Teams that release the implementation resource at go-live spend the next two months in an unplanned scramble, and the corrections made in that scramble are the ones that show up in an audit.

Budget the discovery

The phase most plans omit and the one that determines the timeline.

Most organisations cannot produce a complete list of their own pay rules.

They exist in payroll configuration, collective agreements, individual contracts and one person's memory.

Budget it as a phase, before configuration, with payroll and HR.

Expect contradictions and rules nobody can justify.

Resolving those is an organisational decision on the critical path, not a configuration task.

Plan the stabilisation

The phase that is never in the plan and always happens.

Four to eight weeks after cutover, where corrections are frequent and questions constant.

Someone available, with time allocated.

Teams that release the implementation resource at go-live spend the next two months in an unplanned scramble.

And the corrections made in that scramble are the ones that look wrong in an audit, because they were made quickly and documented poorly.

Schedule integration work explicitly

Project-system connections need configuration, test data and an owner. This implementation example can seed the plan without replacing an end-to-end rehearsal.