IT separation is the workstream that runs the clock.

Why IT separation runs the clock.

IT separation is the longest dependency chain in any TSA exit. Finance cannot close its books on Newco systems until the general ledger is migrated. HR cannot run independent payroll until identity is cut over. Procurement cannot transact on its own contracts until source to pay is hosted on Newco infrastructure. Every workstream waits on IT.

The lead time is also the longest. A mid-market carve-out with two ERPs, a single HRIS, and a standard collaboration suite typically needs 12 to 18 months to fully separate. Larger or more complex separations stretch to 24 months. Anything below 12 months requires either a clean SaaS landscape, an aggressive lift and shift, or a stand-up of replacement applications in parallel with extended dual operation.

The clock starts the day the carve-out is signed. The seller has no incentive to accelerate. Every month of delay is another month of TSA revenue. Buyers who treat IT as a parallel workstream rather than the gating workstream find themselves paying extension fees on every other service catalog item, because everything else is downstream of identity, network, and applications.

The implication is operational. The buyer's IT separation plan is the master schedule. Every other workstream's exit milestone is set by IT. If the IT plan slips, the entire exit slips. Building the master plan as an IT plan, with other workstreams attached, is the only sequence that produces a deliverable exit date.

The five IT subdomains that have to separate.

Identity and access. The first thing to migrate and the last thing to fail cleanly. Every other application authenticates against identity, so until Newco runs its own directory, every other cutover carries a hidden dependency on the seller. Identity separation typically takes 90 to 150 days when starting from a single tenant and longer when the seller's directory is federated across business units.

Network and connectivity. The Newco needs its own carrier circuits, its own firewalls, its own internet egress. Dual stacking is common during transition, with seller VPN access maintained for application reach. The work itself is straightforward. The procurement lead time on carrier circuits is the constraint, often six months for international locations.

End user computing. Laptops, virtual desktops, collaboration tools, mobile devices. Volume work, but it can be parallelised. The complication is timing. End user computing must cut over after identity and before applications, because users need their own credentials to log into the migrated applications. A botched sequence creates a window where employees cannot do their jobs.

Applications. The longest pole. ERPs, HRIS, CRM, payroll, finance reporting, data warehouses. Each application has its own migration approach, its own data conversion challenge, its own integration footprint. A typical mid-market carve-out has 80 to 150 in scope applications, of which 20 to 30 are critical path.

Infrastructure and data centre. Hosting, storage, backup, disaster recovery. Often outsourced to a third-party provider for Newco rather than rebuilt, because rebuilding takes 18 to 24 months and the buyer rarely has the capacity. The decision to rebuild versus outsource is set during the first 60 days post-close and is rarely revisited.

Where the cost lives and why it overruns.

IT separation cost runs in two streams. The first is TSA service fees paid to the seller during the transition. The second is the Newco standup cost, paid to systems integrators, software vendors, and the Newco's own IT team. Most buyers underestimate the second stream by 30 to 50 percent.

The TSA service fee for IT is typically the largest single line item in the agreement. A reference range for a mid-market carve-out is $80K to $600K per month, with the high end driven by ERP hosting, identity, network, and licensing pass-through. Detailed cost ranges by workstream are covered in the TSA exit cost benchmarks.

Standup cost overruns concentrate in three areas. Application migration data conversion, where the data quality in the seller's source system is worse than expected and conversion logic has to be redone. Integration rebuild, where the seller has more undocumented interfaces than the runbook records. Licensing, where the seller's enterprise rate is replaced by the Newco's smaller volume rate at full vendor list price.

The standup budget should be sized at 1.5 to 2.5 times the headline TSA value to land on schedule. Buyers who build the standup budget around the seller's TSA price find themselves underfunded by the time the cutover work starts in earnest. The right benchmark is independent of the TSA price and based on the application count and integration complexity of the target.

Sequencing the cutover. The order matters.

The canonical sequence is identity first, then end user computing, then network independence, then applications in waves, then data centre exit. Each step has a dependency on the prior step. Skipping a step accelerates nothing. It only moves the failure point further down the timeline.

Applications cut over in waves, not all at once. Wave one is the standalone applications with no real time integration to the seller's ERP. Wave two is the integrated applications with one way data flow. Wave three is the core ERP and the HRIS. Wave four is the analytics and reporting layer that sits on top of the migrated systems.

Cutover dates should be set at the start of the program, not adjusted during execution. A drifting date erodes the buyer's leverage with the seller and with internal sponsors. Hold the date and adjust scope. If the data conversion is at risk, cut a wave four item that is not critical and protect the wave three core. Date discipline is what makes the exit deliverable. The full set of milestones is covered in TSA exit milestones, explained.

Dual run periods should be capped at 30 days for any single application. Longer dual runs become permanent dual runs. The seller charges for both environments, and the operational complexity of running two systems exceeds the cost saving from delaying full cutover. Cap the dual run window in the runbook and enforce it.

What the buyer must own from day one.

The Newco CIO role is the single most important hire of the first 30 days. The CIO owns the IT separation plan, the budget, and the relationship with the seller's IT organisation. Without a permanent CIO in place, the separation plan drifts and the seller fills the vacuum with their preferred sequencing. The seller's preferred sequencing is rarely the buyer's preferred sequencing.

The application inventory is the next deliverable. A complete list of every application in scope, the business owner, the integration footprint, the licensing arrangement, the cutover wave. Most sellers will provide a partial list. The buyer's team has to validate it, fill the gaps, and own it. A missing application discovered in month nine of an 18 month TSA is one of the most common reasons for extension.

The governance committee for IT is separate from the program governance committee. It meets weekly, not monthly. The agenda is fixed: cutover progress against plan, blockers requiring seller action, change requests, financial run rate. Without a weekly IT specific governance forum, IT issues surface in monthly program governance after the slip is already locked in.

The exit ramp clauses for IT services should be tighter than the exit ramp for other workstreams. Exit on 30 days notice with no penalty for any service the buyer has stood up. Without the right exit ramp, the buyer continues to pay for redundant services after Newco systems are live. Half of IT TSA overpayment comes from services that should have been turned off months earlier.

More on TSA exit strategy.