Placement, fractional, or rotation: a decision framework

The three shapes look interchangeable from a distance: one ethicist, contracted somehow. They are not interchangeable in what risk they are actually built to cover:

What a placement is actually for

A placement, full-time, one organisation, is built for continuous exposure: a company shipping AI-driven decisions regularly enough that no external cadence could keep pace with the review load. It suits an organisation past the point of occasional oversight, where the role has become a standing function.

What fractional is actually for

Fractional work suits an organisation with real but irregular exposure, a handful of significant deployments a year rather than a constant stream. The ethicist is present to build institutional memory of past findings, but not so embedded that a single organisation is the entire portfolio, which is itself a form of independence: nothing to lose by escalating hard with any one client.

What rotation is actually for

A rotation or cohort placement suits an organisation that wants the seat filled continuously but is comfortable with a practitioner who is still building the specific pattern-matching that comes from seeing many deployments. It also suits the practitioner, a structured way into the field with defined handover points, which is why it doubles as an entry path, covered in our piece on rotation versus pupillage.

The one-question test

Ask how often a new deployment or a significant model change actually ships. Monthly or faster points toward a placement. A handful of times a year points toward fractional. Continuous need paired with tolerance for a practitioner still building depth points toward rotation. The shape follows the cadence of the risk, not the size of the budget.