Migrating an email platform without pausing the campaigns.

Migration work is finite and planned. Campaign operations are continuous and interrupt-driven. Put them on one team and the planned work always loses.

The program was modernizing an enterprise email ecosystem while that ecosystem kept doing its day job: triggered communications going out, journeys running, campaigns landing on a calendar the business had already committed to. Nothing could be paused to make room for the rebuild.

That produces a specific failure mode. Migration tasks have no deadline pressure until the very end, while campaign requests arrive with real dates attached. Left alone, a squad will service the urgent thing every time and the migration quietly slips a week at a time until the final month becomes a crisis.

My mandate covered both halves: run the delivery, and make sure the engagement was shaped so that doing both was actually possible.

The work needed five disciplines sitting together permanently rather than being requisitioned per task. Handing a campaign build to a shared pool would have added a queue in front of every deadline.

A requirements analyst sat alongside those four, because in a bilingual retail operation the cost of a misread requirement is a send that cannot be recalled.

The engagement had been written as a deliverable-based project. The work behaved like a managed service. That mismatch is not a paperwork problem—it decides who absorbs the cost of every change.

A deliverable-based statement of work assumes you can name the outputs in advance and mark them complete. A six-month marketing operations squad cannot: priorities move weekly, a campaign gets pulled, a migration dependency lands late, BAU volume spikes. Under rigid phase-completion language, every one of those normal events becomes a change request, and the team spends its attention on contract administration instead of delivery.

So I worked the commercial structure until it described the actual engagement:

When additional development capacity became necessary partway through, that structure meant it could be added as a resource change against an existing framework rather than renegotiated as new scope.

The migration and the campaign calendar both kept moving, and reprioritization stopped carrying a commercial penalty.

The more durable outcome was the model itself. Once an engagement is described in the terms the work actually operates in, ordinary volatility stops being treated as a deviation—which means it stops consuming the attention of the people delivering.