Spreadsheets or an ERP for an aerospace job shop: how to decide

A tracker sheet is cheap and quick. For a shop with a handful of jobs in flight, a job list, a certificate log and a drawing index can work. Starting on sheets is fine. Plenty of good shops ran for years on a job list, a gauge calibration log and a drawing index kept by a careful quality manager.

The control is the person. One careful quality manager who knows every row, every filename and every supplier habit can hold the whole arrangement together. Those sheets are fine at low volume. They hold a shop together while the paperwork stays small enough for one reader to check line by line against shipped serial numbers.

That is also the weakness.

Ask what happens on their leave. Or when two people open the certificate log at once, and the second save wipes the first person’s entry for heat lot 4471. Watch for the first near miss. It always comes. A wrong revision caught at final inspection, or a certificate missing on the morning of shipment, signals that memory has become the control. Memory fails in holidays and busy weeks.

Customer audits expose the rest. An auditor picks serial 0042 and asks for the drawing revision, heat lot, anodise certificate, CMM report and nonconformance history, all within the hour. That search is too slow. A floor with several copies of the truth is another sign: the foreman’s whiteboard, the planner’s workbook and the quality log each show a different status for one job.

Software will not fix an undefined process. Write down how a job moves from purchase order review to shipment, who signs at each step, and which records must exist. Process comes first and tools second.

List your customers’ flow-down requirements. Some demand set traveller content, AS9102 first article forms, or retention of mill certificates as scanned PDFs, and the system must hold those records. Machine hour tracking alone does not. Include any export-controlled drawings in that list. Who may open a file, and from which location, becomes a real question once records leave a single desktop and travel between the planner, the inspector and the customer.

Clean the data you will migrate, meaning part numbers, revisions, open jobs and approved processors, before anyone enters a single record into the new system.

Choose three real jobs, including one that went wrong, and run them through each option asking whether the system shows revision, heat lot and inspection record for each serial. Score each option on retrieval time, revision control, certificate linking and how it behaves when two people edit the same record at once. A simple table beats a feature brochure.

Ask who can change a released revision and whether the change is logged, then ask how a micrometer’s calibration due date is flagged on the inspection step.

Distrust any demonstration run only on tidy sample data with no rejected parts, no rework routes and no superseded drawings. Tidy demos hide the mess. Ask for a live retrieval in front of you, starting from a serial you choose, and count the screens needed to reach the certificate and the inspection data.

You need not replace everything at once. Many shops begin with revisions and travellers. Certificates come second, and inspection records and calibration follow once the first two hold steady in daily use on the shop floor. Retire sheets one at a time. First the replacement must survive a real audit question, such as the full history of one serial with its heat lot, process certificates, operator confirmations and signing inspector.

Run the new records alongside the sheets for a few weeks and compare serial by serial, retiring a sheet only when its replacement has passed a mock audit retrieval. Budget time for people as well. The planner, the inspector and the operators each enter records differently in a new system and will make mistakes in the first fortnight.

This week, pick one serial and time how long a full record pull takes today, from the drawing revision to the last inspection signature. Write the minutes on the worksheet.