Everyone agrees obsolescence should be monitored. The programs that survive it are the ones where somebody is named, and the review is on a calendar.
Obsolescence management fails in a predictable way. A tool is bought, a report is generated, and nobody owns the follow-up. Two years later a product-change notice arrives about a part that went end-of-life eighteen months earlier and the response is a scramble.
What makes it work is unglamorous ownership. One named person, one review per year per program, one list with three columns: part, status, action. Most lines will read "no change" and that is fine — the value is in the two lines that changed.
The actions themselves are limited and known: qualify an alternate, redesign the affected section, or buy the remaining life. Each has a cost and a lead time, and each becomes cheaper the earlier it is chosen. A last-time buy decided with twelve months of notice is a purchasing exercise; the same decision with two months of notice is a bet.
For products we build, the watch runs from the same bill of materials that drives production, which is the only version that is reliably current. A watch that runs against an old export is monitoring a product that no longer exists.
— GANI Lifecycle engineering team