Most shortage reports are written after the shortage exists. The part is out, the build is stopped, and the report becomes a record of what happened rather than a warning of what is coming. A forecast that is worth the name has to be built from signals that move weeks before the line does, and it has to be read on a fixed day rather than when someone remembers.
Lead time spread, order acknowledgement states and coverage per part number are enough to place every line in one of three windows, and each window has a different action available.
Why shortage reports arrive too late
The first reason is the difference between acknowledgement and confirmation. A purchase order that has been received has a number, a date field and no material behind it. Systems treat it as covered. A supplier who has confirmed a ship date has committed capacity, which is a different fact, and the gap between the two is typically two to six weeks.
The second is that lead times are read as a single number. A part quoted at sixteen weeks that arrives at sixteen weeks most of the time still arrives at twenty two in one case out of ten, and that tenth case is the one that stops a build. Planning on the average is planning for the half of outcomes that never hurt.
The third is allocation. When a maker puts a family on allocation, the notice reaches the customer through a sales contact and the quantity is cut on the next confirmation. Nothing in the planning system records that the supply base changed, so the forecast keeps using last quarter's numbers.
The fourth is scope. A forecast built only from open work orders cannot see the next program, and those parts carry the long tooling or wafer lead time that had to start three months ago.
Signals that move before the line stops
Each of these can be measured weekly from data a purchasing team already holds.
| Signal | Where it shows up | Notice it gives |
|---|---|---|
| Weeks of cover per part number | On hand plus on order, divided by weekly usage | Four to twelve weeks, depending on lead time |
| Past due purchase orders | Any line whose promised date has passed without a ship notice | Immediate, and usually the first real warning |
| Orders acknowledged but not confirmed | Purchase order status field | Two to six weeks |
| Lead time quoted now against lead time at order | Fresh quote from the same channel | One quarter, if the field is refreshed |
| Allocation or cut quantity notices | Confirmation versus ordered quantity | One to two order cycles |
| End of life and product change notices | Maker notification with a last order date | Three to twelve months, the longest warning available |
| Single source lines with no approved alternate | Approved vendor list versus BOM | Permanent exposure, visible today |
The last row is the one most often missed. A part with no approved alternate has no recovery route, so its cover requirement is the lead time plus the qualification time, not the lead time alone.
Three planning windows, three different actions
The mistake in most reviews is applying one response to every shortage. What can be done at twelve weeks is not what can be done at two.
| Window | Question to answer | Action available |
|---|---|---|
| Zero to four weeks | Can the build be resequenced around the missing part | Split the kit, build the boards that do not need it, expedite only what stops shipment |
| One to three months | Is there an approved equivalent, and who signs it off | Cross reference and written customer approval, or a broker route with incoming verification |
| Three to twelve months | Is the part reaching end of life, and what quantity carries the service tail | Last time buy sized on the remaining program life, with storage and shelf life checked |
| Beyond twelve months | Should this part be on the BOM at all | Change the BOM now, while there is still time to qualify a replacement |
A forecast that only reports a number of weeks of cover cannot distinguish these, and the team ends up expediting parts that should have been replaced a quarter earlier.
How SUPERB PCBA keeps a shortage forecast current
Every kit is counted by X-ray against the reel or tray on receipt rather than trusted by label, so the on hand figure in any coverage calculation is a measured quantity. Quantity, date code and maker markings are compared before the count is accepted, and parts with provenance risk are opened and checked.
Shortages are reported against the build date rather than the order date, with the reason attached: no stock, order past due, allocation cut or lead time moved. That single field is what lets a customer decide between expediting and substituting without a meeting.
Where an alternate exists, a pin to pin comparison is prepared and sent for written approval before anything is bought or fitted. Where no alternate exists, the last time buy quantity is calculated against the program tail and the storage the part needs. The assembly side backs this with seven SMT lines, 01005 placement, 3D SPI and AOI, X-ray at IPC-7095 with voiding under five percent, and seventeen process steps with ten QC stations under IPC-A-610 Class 2 and Class 3, ISO 9001 and ISO 13485. Components carry a 730 day warranty.
What a forecast cannot fix
It cannot create capacity. When a wafer fab is fully booked, the only levers are the buy position and the margin already committed, and both were decided months earlier.
It cannot correct a BOM that is wrong. A forecast built on the wrong part number is confident and useless, which is why the BOM scrub comes before the coverage report rather than after it.
Component shortage forecast questions
How far ahead is a forecast useful?
Twelve months for parts with long tooling or wafer lead times, three months for everything else. Beyond a year, demand assumptions move enough that the exercise turns into a scenario review rather than a plan.
Should the forecast include parts not yet ordered?
Yes, at least for single source and long lead items, because those are the parts where the order has to be placed before the build is released. The follow on work sits under safety stock planning.
What is the first thing to fix if nothing is measured today?
Past due purchase orders. It needs no new data, it is available in every system, and it catches most shortages that are already in progress. From there, add weeks of cover per part and the acknowledgement state, then the last time buy and LTB planning review for anything with an end of life notice.
Next step
Send the BOM with twelve months of usage per line and the open order status if you have it. The reply returns a coverage view per part with the reason for each gap, the lines with no approved alternate, and the parts where a cross reference would remove the exposure. Where the gap is already open, the sequence is in BOM shortage recovery and the lead time side in component lead time solutions.