Third and last of the three irreversible decisions.
Changing the employment model after work relationships exist is not a configuration change. It is a data migration, and it has statutory consequences for service dates — which is where legal entitlements attach. Treat it as irreversible. Not "difficult". Irreversible.
The default recommendation is single assignment, unless you have a concrete case that needs more. And the reason is not technical: simplicity has real operational value. Every additional layer is one more thing every HR user navigates, every day, forever — and one more thing for them to get wrong. That cost is invisible on a design document and extremely visible in a support queue.
The questions that force multiple assignments: does anyone do two different jobs for the same employer, concurrently, with separate pay or reporting? Are there bank, casual or relief workers who work across departments? Does the client operate a shift system where one person holds two rostered roles? And the ones that force the intermediate layer: is there a contract or terms document governing multiple assignments together? Do statutory obligations attach at a level above the individual assignment? If the answer to everything is no, take the simple model and stop.
Now the conversation you will actually have. Clients frequently ask for the most flexible model, just in case — it sounds prudent and it looks free. Push back. The flexible model adds a layer every HR user navigates every day, forever, to serve a case that may never arise. Get a concrete example, or choose the simpler model: not a hypothetical, but a person, a department, a situation somebody can name. If nobody can produce one, that is your answer. Have that argument in week four rather than letting the client discover in year two that they are paying a daily tax for a case that never happened.
Mixed models by legal employer are the honest middle — the healthcare subsidiary needs multiple assignments, the head office does not. Be straight about the cost: two models means two sets of training and two answers to every question. That is real, and it is still usually cheaper than putting everyone on the complex model.
And put the choice in the decision log with the specific business case that drove it. This decision gets questioned more than any other — more than the LDGs, more than jobs versus positions — because it is the one HR users bump into daily.