Before the money moves. The structure, the ownership percentages and the escrow terms are all evidence, and they are far easier to arrange correctly than to repair. Deals that close first and consider immigration afterwards produce files where the capital is spent but the record does not prove it was at risk.
Sequence the deal so the evidence exists by closing
Hypothetical example: Colette has agreed in principle to buy an American self-storage facility. The natural instinct is to complete a purchase and then apply, which puts every immigration question into the past tense. Working the other way round costs little.
Ownership percentages, the shareholders agreement and the management structure can be settled during negotiation, when the other side is still willing to accommodate them. Escrow can be drafted so that funds are irrevocably committed while still protecting Colette if the application fails, which is a drafting problem best solved before signature. Due diligence can be directed at the documents the application will need, so the same accountant's work serves both purposes.
Where the business is being built rather than bought, spending sequenced properly counts toward the investment, while informal advances made before any structure existed can be difficult to attribute. None of this is urgent until it is retrospective. The one date nobody can move retrospectively is the day funds leave an account, which makes it the date the rest of the plan should be arranged around.