Time per iteration spent searching for the right file, confirming which version is current, or asking a colleague — before any design work starts.

With ownership, maturity state, and one governed location (e.g. 3DEXPERIENCE), the correct current document is just there. This alone doesn't need AI — it's the prerequisite the rest of the model builds on.

This isn't mostly individual mistakes — it's inter-engineer variance. Ten engineers given the same spec produce ten slightly different drawings, and every one has to be re-interpreted downstream, in manufacturing, procurement, and QA. Encoded rules make the same rule produce the same drawing every time — consistency, not just speed. That's Lean applied in engineering, before it reaches the shop floor.

A dimensional error caught at fabrication — scrap, remake, schedule slip — not one caught in design review.

The share of iterations your encoded rules can't handle, routed back to manual work at today's rate. Often the most underestimated number in the whole model — start at 0 and see how sensitive your case is.

The setup cost (F) is paid once and amortizes across every project that reuses the same encoded rules.