The Roles You Can’t Replace Mid-Program
Most SAP staffing plans are built as headcount. So many consultants, so many months, mapped against workstreams. That is the format the budget wants, and it is the format everyone produces.
Headcount and continuity are not the same thing, though, and only one of them appears in the plan.
Where it actually breaks
Every conversion program has roles where turnover is an inconvenience and roles where turnover is a re-do. The line between the two rarely gets written down, which is why it usually gets discovered somewhere in the second data migration cycle.
The roles that carry real continuity risk tend to share one trait. They hold context that was never fully documented, because documenting it properly would have cost more than doing the work.
The data migration lead is the clearest example. By cycle two, that person is carrying a mental map of which legacy records are genuinely bad, which are bad-but-load-anyway, which business owner signed off on which exception, and which mapping rule exists because of a decision made in 2013 that nobody can explain any more. Hand that over and you do not lose a person. You lose two cycles while the replacement rebuilds the same map.
Cutover managers are similar. The value is not the plan document — anyone can inherit a plan. It is knowing which of the 400 steps have slack, which teams routinely over-estimate their duration, which interface fails on first attempt in every rehearsal, and who to call at 3am when it does.
Then there are the config leads in whichever modules your estate has diverged furthest in. Usually FI/CO, often MM or SD, sometimes something specific to your industry. They are carrying the reasoning behind decisions that look arbitrary from outside and are load-bearing on the inside.
What teams get wrong
The common mistake is backfilling by role title. A senior data migration lead leaves, a senior data migration lead is brought in. On paper the plan is intact. The gap does not appear in the staffing report at all.
It appears about six weeks later, when decisions that were settled start getting reopened. Not because the new person is weaker — often they are stronger — but because they cannot inherit judgment that was never written down, and their instinct is to re-verify rather than accept.
That is the right instinct. It is also two cycles.
The second mistake is treating a start date as a mobilisation date. Someone joining in week one of a workstream and someone joining in week nine are doing different jobs, even with the same title. The person joining in week nine spends their first month reconstructing context instead of producing.
What tends to work better
A few things hold up consistently across programs.
Name your continuity roles explicitly, before the program starts. Not “senior resources” — the specific three or four seats where a mid-program change costs you cycles. Everyone on the program should know which ones they are.
Build overlap into those seats deliberately. A second person close enough to the work to take over without a rebuild costs less than the rebuild does. It is a hard line to defend in a budget conversation and it is usually the right one.
Keep a decision log that captures reasoning, not just outcomes. “We excluded these material records” is an outcome. “We excluded them because the plant closed in 2019 and finance confirmed no open items” is the thing that stops the question being reopened in cycle three.
And where you can, secure those seats earlier than the rest of the team. Not because they are more senior, but because they are the ones you cannot swap.
The honest constraint
None of this is free, and some of it is hard to justify against a plan that measures headcount. But the trade is straightforward. You either pay a little for continuity up front, or you pay for it in cycle two, in rehearsal, and eventually in the cutover window where there is no slack left to absorb it.
Plans that treat every seat as replaceable are easier to build. They are just less likely to survive contact with the second data load.