Separation work
The deal closed. Now the platform has to come apart.
Separating a divested business out of a shared ServiceNow instance, against a transition services agreement that expires whether or not you are ready. Consolidations slip. Separations do not get to.
Why this is not a migration
A separation is a different problem wearing similar clothes
The tooling overlaps. The failure modes do not. If you scope a carve-out as a migration run backwards, you will find the differences late, which on a TSA clock is the expensive way to find them.
You are subtracting, not adding
A consolidation asks what moves. A separation asks what leaves, and every record you do not explicitly decide about stays where it is by default. That default is the whole risk: the divested business arrives missing history nobody thought to claim, or the parent keeps data it no longer has any right to hold.
The date came from a deal, not from IT
Nobody in the platform team chose it and nobody in the platform team can move it. It was set when the transaction closed, and the transition services agreement is what makes it real: the parent is contractually obliged to keep serving the divested business for a fixed window, and paying for it.
Every day past the deadline costs money
A slipped consolidation is embarrassing. A slipped separation extends a TSA, and TSA extensions are negotiated from a weak position because both sides already agreed the original term was enough. That is why this work gets funded properly and why the date holds.
Two live platforms, both in production
You cannot pause either side. The parent keeps running its business and the divested entity keeps running its own, often on the same instance, for months, while you pull them apart underneath. Consolidations get a quiet weekend. Separations rarely do.
How we run it
Four moves, in this order
Principal-led, which is what makes the order keepable. One senior engineer who holds the whole picture, rather than a handoff between a discovery team and a build team.
-
First
Decide who owns what, on paper
Not a technical question and not one the platform team can answer alone. Which accounts, which contacts, which historical tickets, which knowledge articles belong to the business being sold. Legal and the deal team have opinions here and they need to be captured before anyone writes an extract, because this is the step that gets relitigated at the worst possible moment.
-
Then
Find what the org chart does not show
Shared catalog items, approval chains that route through a manager who is leaving with the divested entity, integrations authenticating as a service account nobody has owned in years, reports that quietly span both sides. These are what turn a clean split into a six-week argument, and they are all findable before you start.
-
Then
Rehearse the split at real volume
Build it re-runnable and run the whole separation end to end in a non-production environment against production-sized data. What a rehearsal finds is worth more than what a plan predicts, and on a TSA clock it is the only affordable place to find it.
-
Finally
Cut, and prove both sides are whole
A runbook with owners, blackout windows and a rollback path. Afterwards, reconciliation has to answer two questions rather than one: that everything which should have moved did, and that everything which should not have moved did not. The second is the one people forget, and it is the one with legal consequences.
What we have actually done
Being straight about this, because it matters more than a claim would: our published work is consolidation rather than separation. We have migrated more than fifty million records between ServiceNow instances in a single weekend cutover with none lost, and folded an entire support operation off three legacy platforms into one instance against a go-live date the business set. The case studies have the numbers.
A separation runs the same machinery in the other direction: the same profiling, the same signed-off mapping, the same re-runnable pipeline, the same record-by-record proof. The difference is what the reconciliation has to prove, and the fact that the clock belongs to a contract rather than to a plan.
If you want a firm that has done your exact deal before, that is a fair thing to want and we will tell you plainly whether we are it.
Booking Q4 2026 and Q1 2027
Separation work wants its ownership decisions and its dependency map done months before the TSA expires, not weeks. If your transition window closes in the first half of 2027, the useful conversation is now rather than in the new year.
Tell us what is being separated, roughly when the TSA runs out, and whether both sides are on the same instance today.