Five lessons that repeat across local and global rollouts
Different countries, different systems, the same failure patterns. The lessons that consistently change the outcome of a complex rollout.
Sunny Kaul
Principal Advisor, Kaul Advisors
Every rollout has a context that makes it feel unique: the countries involved, the legacy landscape, the maturity of the teams, the regulatory environment. Those differences are real. They are also why the same few lessons keep resurfacing.
These are the patterns we return to when a programme is under pressure.
1. Local complexity is not a detail
Global templates create a useful baseline. They can also create the illusion that local variation is a minor configuration exercise. It rarely is.
Statutory reporting, local payroll rules, tax requirements, languages, and established workarounds can each reshape a rollout wave. The earlier the local process owners are involved, the fewer surprises arrive during testing.
A localisation register is a simple control: one place that records each local difference, its owner, its impact, and the decision on how to handle it. It should be reviewed alongside the plan, not discovered during cutover.
2. Data readiness is a programme risk
Data remediation is routinely planned as a downstream activity. It then becomes the reason testing slips and cutover pressure rises.
The pattern is consistent because data work is hard to estimate and easy to defer. Ownership is often unclear, quality thresholds are vague, and the programme assumes the business will produce clean data when asked.
We prefer to treat data as a phase gate. Before a wave proceeds, the programme should be able to show:
- named data owners for each priority domain
- measured quality against agreed thresholds
- a migration rehearsal with realistic volumes
- a documented plan for remaining exceptions
This is not bureaucracy. It is the difference between a cutover that is planned and one that is hoped for.
3. Governance should shorten decisions
A steering committee that reviews a long status pack without resolving anything is an expensive audience. Governance exists to make decisions, not to observe activity.
The most effective governance we see has a short agenda built around:
- decisions required
- risks that changed
- dependencies needing escalation
- scope or sequencing choices
Operational issues should be resolved below that level. If they are not, the steering committee becomes a bottleneck and senior attention is spent on work that has a clear owner elsewhere.
A programme does not need more governance. It needs clearer decision rights and a shorter path to the people who can exercise them.
4. Pace must match absorption capacity
A rollout can be technically ready and organisationally premature. Training may be complete, but managers may not yet be confident. Support may be staffed, but the knowledge may still sit with the programme team.
Readiness criteria that rely only on testing completion will miss this. We include observable adoption signals: process adherence, data quality at entry, first-line resolution rates, and the quality of questions coming into support.
Slowing one wave can feel like failure. Often it protects the credibility of the whole programme.
5. The first close is part of the rollout
Go-live is a milestone, not an ending. The first financial close, first payroll, and first peak period are where many programmes discover whether the operating model actually works.
These events deserve their own plan: who is on point, what is monitored, how issues are triaged, and what “stable” means. Hypercare should have explicit exit criteria and a permanent owner named before it begins.
The common thread
None of these lessons are about technology. They are about making the current state visible early enough to act, giving decisions a clear owner, and resisting the temptation to treat the plan as evidence of readiness.
The specifics vary by country and programme. The discipline does not.
Keep reading
Related articles
PM Lessons Learned The quiet cost of skipping change management
A system can go live on time and still fail to land. The cost of underinvesting in adoption rarely appears in the status report — until it does.
Read article
KA Labs Designing a cutover runbook that teams actually use
A cutover runbook is not a record of intent. It is the decision tool a tired team relies on at 3 a.m. Here is how to write one that holds up.
Read article
KA Labs What ERP assessments get wrong about process discovery
Discovery is not a tour of every process. It is a focused inquiry into the few decisions that will shape fit, risk, and the credibility of the programme.
Read article