Knowledge That Walks: Making Manufacturing Data a Condition of Award

A buyer issues a drawing, receives a part, signs off the inspection report and files the paperwork. On the face of it, the transaction is complete. The drawing remains in the buyer's document control system, the purchase order is closed, and the component can, in theory, be sourced again. In practice, that theory often fails at the first change of supplier. The part can be re-made only if someone can reconstruct the manufacturing route, and that route is rarely contained in the drawing pack.

This is not a comment on supplier goodwill. It is a structural gap in how precision subcontract work is bought. The drawing defines the destination. The CAM programme, the post-processor settings, the fixture design, the probing cycles and the setup sheets define the route. When the route lives only with the machinist who built it, the buyer owns the specification but not the means of production.

The drawing is the destination, not the route

A drawing is a contractual description of a finished component. It says nothing definitive about stock selection, ordering of operations, workholding, tool engagement, thermal compensation or in-process verification. Those decisions are made during process planning, and they are encoded in artefacts that sit outside the drawing frame.

In a simple job, the gap may be bridgeable. A competent machinist can re-create a route from the drawing and a sample part. In a complex job, the gap is wider. A five-axis programme with a carefully developed post-processor, a bespoke fixture with known deflection characteristics, and a probing routine that corrects for machine geometry are not generic assets. They are the accumulated product of trial, measurement and adjustment. They are also the reason the part can be made to tolerance at a predictable cost.

What actually travels with a job

When a precision job moves between suppliers, or when a buyer seeks a second source, several categories of data matter. The first is the CAM source file, not a neutralised model or a screen capture. The second is the post-processor configuration and the machine model it was written for. A programme that runs on one machine may not run on another without re-posting, and a post-processor is a piece of intellectual property in its own right.

The third category is fixture design: CAD models, assembly drawings, photographs of the as-built fixture, and notes on wear parts or adjustment points. The fourth is the probing and in-process measurement routine, including the nominal values, tolerances and logic used to decide whether a cut proceeds. The fifth is the setup sheet and tool list, with the caveats that experienced machinists attach to them.

None of this is exotic. It is ordinary manufacturing documentation. The problem is that it is rarely listed as a deliverable in the purchase order.

Why it is rarely in the terms

Commercial discussions tend to concentrate on price, lead time, quality approvals and payment terms. Manufacturing data is treated as an operational matter, something the supplier will obviously retain because the supplier needs it to make the part. That is true while the supplier remains the supplier. It becomes less true when the relationship ends, when capacity is moved, or when a buyer wants to test the market.

There is also an IP sensitivity. Some suppliers regard CAM programmes and fixture designs as their own competitive advantage, and they are not wrong to do so. A buyer who asks for them without discussing ownership, licensing and confidentiality is asking for a commercial argument rather than a technical deliverable. The answer is not to avoid the request. It is to define what is being bought, on what terms, and at what point in the relationship.

The machine-specific trap

A common assumption is that manufacturing data is portable. It is portable in principle and conditional in practice. A CAM programme is written against a machine, a control, a tooling set and a post-processor. Move it to a different machine and it may require re-posting, re-validation and fresh prove-out. A fixture may be designed around a particular vice, pallet or tombstone. A probing routine may depend on the kinematic model of a specific machine.

This matters for procurement because it changes what 'handover' should mean. The buyer is not simply acquiring a file. The buyer is acquiring enough information to reconstruct a validated process, or to hand a new supplier a credible starting point. That distinction should shape the specification. A neutralised model is better than nothing, but it is not a route. A programme without its post-processor and machine assumptions is a partial asset.

Writing handover into the award

The practical remedy is to make manufacturing data a condition of award rather than a negotiation at exit. At tender stage, the buyer can request a schedule of deliverables: CAM source files, post-processor details, fixture CAD and photographs, probing routines, setup sheets and tool lists. The schedule can distinguish between data held in escrow, data delivered at project close, and data delivered on request during the contract.

Ownership and licensing should be settled in the same conversation. The buyer may not need exclusive rights to a supplier's generic post-processor, but it may need a perpetual licence to use project-specific data for continuity of supply. Confidentiality obligations should run in both directions. The aim is not to strip a supplier of its know-how. It is to ensure that the buyer is not left with a drawing and no route when the supply relationship changes.

A practical schedule

A workable schedule need not be elaborate. It should identify the data items, the format, the revision status and the trigger for delivery. It should state whether the data is to be validated against the as-built part and, if so, by whom. It should record the machine assumptions and any known limitations. It should also name the person responsible for maintaining the data when a programme is revised.

For UK buyers, this is particularly relevant where a component has a long service life or a safety-related application. The cost of reconstructing a process from scratch is rarely visible at the point of award. It appears later, as engineering time, prove-out scrap and delivery risk. A modest documentation requirement at the start is usually cheaper than a urgent data request at the end.

The uncomfortable conclusion

A part that cannot be re-programmed is a part that is only partly controlled. The buyer may own the drawing, the specification and the inspection report, but not the means to reproduce the component within a predictable cost and lead time. That is a procurement risk, and it belongs in the commercial terms.

The drawing pack is necessary. It is not sufficient. Precision supply is a chain of decisions, and the buyer who wants resilience should buy visibility of those decisions, not just the output. The toolpath may not belong in the purchase order, but the obligation to hand it over should.