Six ways TSA exits fail to deliver their business case

A transition services agreement is supposed to buy time and then end, releasing the cost that sat behind it. In practice a large share of TSA programs reach their final exit having delivered almost none of the savings that justified them.

The failure is rarely dramatic. Nobody misses a date badly. Services get exited more or less on schedule. And then finance runs the numbers and the run-rate has barely moved.

Six mechanisms produce that outcome. All six are visible early, and all six are preventable.

1. The service was exited but never decommissioned

This is the largest single cause, and it is almost always a sequencing oversight rather than a decision.

Exit certification confirms the recipient no longer needs the service. It does not terminate the provider's underlying capability. The platform stays licensed, the infrastructure stays powered, the support contract renews on its own schedule, and the staff who delivered the service stay in post because nobody told them otherwise.

The fix: treat decommissioning as a required step of the exit, not a follow-on activity. Track it per service through infrastructure release, licence termination, support contract termination, and asset record update — and require finance to verify the cost has left before the exit is considered closed.

2. Exit planning started after close

On Day 1 everyone is exhausted and focused on stabilization. Through months two to six the recipient is building an organization and firefighting. By the time anyone turns to exit planning, the notice period is approaching and there is no longer time to build the replacement capability.

The fix: exit plans are a Day 1 readiness deliverable. Every service should have a named exit path, an owner, a budget, and a target date before the first invoice is raised.

3. Exit dates were set service by service

Services do not exit independently. An application exit may require an identity exit first, which requires a network exit, which depends on a platform decision nobody has made.

We have repeatedly seen a single unmade decision gate more than half a TSA portfolio. Programs that set dates service by service, without mapping the network, produce a schedule that collapses on contact.

The fix: build the exit dependency network before setting any individual date. The critical path through that network is the real program duration.

4. The target exit date was the contractual end date

Setting the target at the contract end leaves no room for a slipped milestone, a vendor delay, or a failed cutover. There will be one of those.

The fix: set targets inside the term with genuine buffer, and red-flag any exit that is inside its notice period minus sixty days with less than eighty percent of the plan complete.

5. There was no named TSA manager on each side

This single omission degrades everything downstream. Without a named accountable individual on both the provider and recipient side, TSAs drift into informal service delivery, undocumented scope changes, and disputed invoices. By the time anyone attempts to reconstruct what was delivered, the evidence is gone.

The fix: name both managers in the schedule itself, before Day 1, and stand up service and cost reporting from the first month rather than the third.

6. The extension was requested inside the notice period

Extension pricing typically steps up to 110–150% of base, escalating further on a second extension. That step-up exists deliberately, to create exit incentive, and it works.

A recipient who raises an extension inside the notice period has no negotiating position and no alternative. A recipient who raises it ninety days out has both.

The fix: review exit health monthly against notice dates, and escalate anything at risk while there is still room to negotiate.

What good looks like

A well-run TSA program has four things visible at every steering meeting: a burndown of live services against the exit roadmap, the dependency network with its critical path, the count of services exited but not yet decommissioned, and cost eliminated split between claimed and finance-verified.

When the burndown flattens, extensions are coming — and that is visible roughly six months before it becomes a negotiation.