Mostly engineering and records time. Government charges are in the published fee schedule; the substantial cost is extracting and explaining evidence that was never organised for an outside reader.
Ask engineering for an hours estimate
Confirm government charges against the current fee schedule when the petition is prepared and treat premium processing as a separate decision. Then ask the engineering lead for an honest estimate of the hours needed to pull revision histories, locate commissioning and purchase records, and write a technical description a non-specialist can follow without losing the point. That number is usually larger than expected and it competes with billable work.
Add the cost of a professional review that tests the claim against the legal standard before filing, which is where a weak case is cheapest to discover. Where documentation proves genuinely absent, consider whether improving record-keeping now is worth doing before any transfer is contemplated. Hypothetical example: a precision optical-coating manufacturer wants to place the employee who developed its internal yield-analysis method at an affiliated plant.
The first review should list the cost of retrieving archived engineering files, securing confidential records, translating technical material if needed, and paying for a review that tests the specialised-knowledge claim. Repository logs and archived project folders settle whether the evidence exists; finance records settle whether retrieval work has a budget. Do this before promising a filing date to the plant.
The common loss is spending heavily on a narrative before discovering that the decisive documents were overwritten or belong to a third party. A staged review identifies that risk early and can support a decision to preserve records first.