A cable network group, cut loose from its parent's payroll

The spin-off was announced. Payroll still belonged to somebody else.

A national cable network group was being separated from its parent company into a standalone business. The corporate work of a separation gets the attention: the filings, the balance sheet, the announcement. What gets discovered later is that the new company does not own the systems that pay its people.

HR ran on the parent's PeopleSoft instance. Payroll, time, recruiting and employee self-service all ran on platforms the parent held the contracts for. None of it was coming along automatically. The new company had a date by which it had to be off all of it, and that date was set by agreement between two corporations, not by anybody's project plan.

Two programs wearing one name

On paper this was a Workday HCM and payroll implementation. In practice it was two programs running against the same deadline.

The first was the one everybody recognizes: design, configure, convert, test, cut over, go live. The second was the separation itself, and it is the one that decides whether the first one survives contact with reality. Every system the parent was running had to be individually unwound: what data comes across, what stays behind, who can still reach it after the date, for how long, and who answers the phone when someone needs a record from before the split.

The separation matrix

The artifact that held the whole thing together was a separation and cutover matrix. It went application by application and department by department, and for each one it answered a short list of unglamorous questions.

What access continues past the separation date, and until when. What data and reports have to be pulled before the cutoff, and who owns pulling them. What happens to historical payroll and check history. Who at the parent company fields an ad hoc request afterward, and under what arrangement. Which records have a retention obligation that outlives the relationship.

Read the answers together and you get an unsentimental picture of everything a parent company quietly does for a subsidiary that nobody ever wrote down. HR needed read-only access to the legacy system to pull employee history, and only until the parent's license expired. Recruiting needed candidate records extracted before the door closed. Payroll needed check history and balances converted, with a fallback if the history could not come across cleanly. Employment verification, benefits and retirement feeds all had to be re-pointed to a platform that did not exist yet.

It is never the conversion that gets you. It is the fifty things the parent still quietly does for you that nobody wrote down.

Running both tracks against a date that would not move

I ran the integrated HR and payroll plan across both tracks, with the software vendors, the implementation partners and the client's own HR, payroll, benefits and IT teams working to one schedule.

The mechanics were conventional and the discipline was not. End-to-end and user acceptance testing across the full employee lifecycle. Parallel payroll testing, because the only acceptable evidence that a new payroll works is that it produces the same answers as the old one before you retire the old one. An interface log tracking every downstream feed to retirement providers, health carriers and employment verification services, each of which had its own file format, its own trailer counts and its own idea of what a valid record looked like. A benefit provider punch list. Master file and balance conversion. Separate go-live checklists for HCM and for payroll, because they fail in different ways.

Above that sat the governance: an executive steering committee, a weekly reporting rhythm, and five workstreams whose dependencies I owned. The steering committee's job on a separation program is narrow and important. The date does not move, so when something slips, the only lever is scope, and somebody with authority has to pull it quickly and on the record.

What made it hard

Three things, none of which show up in a project plan template.

The deadline was contractual. A normal program that runs late negotiates a new date. A separation program that runs late is a breach, or it is an expensive extension paid to the company you are trying to leave. That changes how you sequence work and how early you force decisions.

The counterparty was also the outgoing parent. Every request for data, access or support went to an organization that was itself reorganizing, with its own priorities and its own people heading for the exits. Goodwill was real but finite, and it decayed as the date approached. Anything you needed from them, you needed early.

And payroll is unforgiving. Most enterprise systems can absorb a rough first week. Payroll cannot. Everyone in the company finds out at the same time, and in a business built on talent and union agreements, a bad first cycle is not an IT problem, it is a labor problem.

The outcome

The company went live on its own HR and payroll platform on time and under budget, with the legacy estate retired against the separation date and no missed payroll.

The measure that mattered was not the go-live date. It was that on the first ordinary Friday after the split, people got paid the right amount, benefits deductions were right, the retirement feed landed, and nobody outside the program noticed anything had happened. On a separation, being uneventful is the whole objective.

Why this matters if you are facing one

Corporate separations create a specific and repeating problem. A company that has never owned its own HR and payroll suddenly has to, on a date it did not choose, while its former parent's willingness to help is on a timer. The Workday work is the visible part and the well-understood part. The separation work is where these programs actually get lost, because it is nobody's obvious job and it does not surface until someone asks a question that starts with "wait, who has the".

Build the separation matrix early, force the access and data decisions while goodwill is still high, prove payroll in parallel before you retire anything, and put one person on the client's side of the table who owns the whole thing and can say no. That is the difference between a separation that ends in a clean first payroll and one that ends in a very public bad Friday.

How I run a separation: the full approach →

Client anonymized. Details available under NDA.