Purchasing and engineering: the same data, a different conversation

Purchasing and engineering look at the same product and see different things. The engineer sees an assembly: dimensions, materials, tolerances and a satisfying bit of puzzle work. The buyer sees articles: numbers, suppliers, lead times and price breaks. Both views are correct, and yet in many companies the two departments mostly talk past each other, usually at the exact moment things are already urgent.

That conversation can change, and the key is not a reorganisation but data both sides can trust. This article is about purchasing and engineering collaboration: where it grinds, what changes once both departments work from the same reliable source, and how supplier knowledge and standard sizes find their way back to the drawing board. A toolbox like Thundercad plays a modest but practical part in that: making sure data from the CAD model reaches purchasing without manual work.

Two disciplines, two languages

For engineering a part is done when it is right: function, strength, manufacturability. For purchasing the work only starts at that point: is this an existing article or a new one, who supplies it, what is the lead time, and is there a price break that makes a slightly larger quantity cheaper per piece? A three millimetre sheet is a thickness to the engineer; to the buyer it is a trade size that is either common or not.

Neither view is more complete than the other; they are meant to be connected. The connection is data: a bill of materials in which every part has not just a name and a size, but also an article number, a material and a marker saying whether it is a make part, a buy part or a standard part. How to keep those three cleanly apart is something we covered earlier in Make parts, buy parts and standard parts: cleanly separated in your BOM.

Three friction points everyone recognises

Ask buyers and engineers where things go wrong between their departments, and you hear the same three stories:

• The BOM arrives too late. Purchasing only gets the list once the design is completely finished. Long lead time items then surface too late by definition, and either the schedule slips or orders go out as rush jobs.
• The specification is incomplete. A line like "motor, 3 kW" without an article number or supplier preference means one thing for purchasing: chasing. Every incomplete line costs an email exchange and a day of waiting, times forty lines per project.
• Changes after the order. Engineering adjusts a dimension, but the order has already been placed. If nobody knows exactly which version was ordered, the deviation only comes to light at incoming inspection, or worse: during assembly.

The three share one root: the information exists before it is shared. The model has known for weeks which parts are in the design; purchasing only hears about it once someone finds time to make a list.

What changes when the data is shared

Turn it around: purchasing looks along as soon as the design takes shape, using a preliminary BOM that comes straight from the model. Long lead times then surface at a moment when something can still be done about them. Order the critical parts from that preliminary list, and let the rest follow at release.

The condition is that the list is trustworthy, not a retyped snapshot. With Export BOM from Thundercad, engineering puts the current BOM into an Excel template in one action, with exactly the columns purchasing needs: article number, description, quantity, material. No retyping, so no typos and no "which version is this anyway". How that route continues technically, from model to ERP, is covered in One source of truth: your BOM into purchasing and ERP without retyping; here we care about what that reliability does to the conversation.

Because that is the real shift: when the list is right, the meeting is no longer about the list. Nobody has to check whether line fourteen is still current. The conversation is about substance: which supplier, which alternative, which order sequence.

Supplier knowledge back to the drawing board

Shared data works in both directions. Purchasing knows things engineering could use every single week: which supplier is good at which operations, which sheet thicknesses and profiles are common, where a price break sits just above the planned quantity. As long as that knowledge lives only in the buyer's head, it lands on the table after release, when changing things is expensive.

In practice, feeding it back means: a list of preferred articles and preferred suppliers that engineering can consult, standard sizes set as the first choice in templates and libraries, and the agreement that a new article is only created after a quick check whether an existing one does the job. The engineer who picks from common sizes designs just as fast; purchasing orders without surprises.

The new conversation in four fixed moments

Collaboration stays vague as long as it depends on goodwill. So anchor it to four fixed moments in every project:

1. At design start. Purchasing names the risk articles: long lead times, busy suppliers, scarce materials. Engineering then knows where design choices can buy time.
2. Halfway through the design. Engineering shares a preliminary BOM from the model. Purchasing places the critical orders and checks specifications for completeness.
3. At release. The final list replaces the preliminary one, with a clear marker of what changed. Purchasing orders the rest.
4. At every change after that. First check what has already been ordered, then change. The buyer hears it first, not last.

Four moments, no heavy procedure. But with one firm agreement underneath: the list comes from the model at every moment, so nobody works from an outdated copy.

Frequently asked questions

Does purchasing need to work with Inventor?

No. Purchasing never has to open a model; the point is that the BOM from the model arrives in a form purchasing uses daily, usually Excel or the ERP system. That translation step belongs to engineering or work preparation, and it is exactly the step you want to automate.

When should purchasing join a new project?

Earlier than feels natural: at design start, when nothing is fixed yet. At that point purchasing only needs to flag which parts will be tricky in terms of lead time or availability. That quarter of an hour often decides whether the schedule is realistic.

How do you prevent purchasing from ordering against an old BOM?

By never letting lists float around as loose copies, but pulling them fresh from the model at any moment. With Export BOM that is one action, so there is no reason to keep an old list around. You can try it without obligation in the free trial month of Thundercad.