Well before the destination project needs the person. Records assembly and any repair to the qualifying year set the schedule, and neither responds to pressure at the end.
Start before the project is committed
A request that arrives with work already underway creates pressure to file a thin case, which is the worst outcome available. Ask the project sponsor when the technician is genuinely needed, then work backwards through the tasks that cannot be compressed: confirming ownership between the organisations, verifying the continuous qualifying year against payroll as it actually ran, and assembling contemporaneous technical records from systems that may be archived. Where the destination operation is itself new, a new-office petition is normally approved initially for one year and requires evidence of premises and capacity to support the role, which creates a separate lead time.
Setting the project schedule around these, rather than discovering them inside it, is the difference between a considered case and a rushed one. Hypothetical example: a museum climate-control systems firm needs its commissioning specialist in the United States for a gallery retrofit. The first review should produce a schedule for confirming the related entities, verifying the qualifying year, collecting technical evidence, and checking whether the destination operation needs new-office support.
Corporate records, payroll, and dated project materials settle those points in that order. Begin archive retrieval before the installation calendar is fixed, because older files may require access approval. What goes wrong is filing from current recollection while the original commissioning data remain on an obsolete server.
The project schedule should include time to locate, authenticate, and explain the records that distinguish internal expertise from a broadly available trade skill.