A project team freezes the design of a custom housing. Engineering releases the drawing with a named material, a tolerance band that only one shop can hold, and a geometry that fits one supplier's existing tooling. Then the request for quotation goes to procurement. By then the part is real — and so is its cost.
An engineered category is not a catalog you shop. It is a string of design decisions you were not in the room to make.
That is the whole problem with engineered and custom components. In a standard part, the spec is shared, the supplier pool is open, and procurement still has choices when the RFQ lands. In an engineered category — a custom casting, a molded housing, a machined bracket, a sub-assembly with its own architecture — the cost is built into the geometry. The material grade, the tolerance, the single-source tooling, even the supplier shortlist implied by the way the part was drawn: all of it is decided before sourcing opens. McKinsey's should-cost work frames the development timeline as a funnel — planning, concept design, detail design, quoting, sourcing, production — and is explicit that the further downstream you go, the more decisions you have already accepted. By the time procurement sees the RFQ, most of the cost has been committed upstream, in engineering.
What actually gets locked.
Three things get fixed in the engineering phase, and procurement can only influence them before the drawing is released.
1. The cost. A custom part's price is not discovered at RFQ; it is the sum of choices made months earlier — material selection, manufacturing route, tolerance, yield. Once those are frozen, the quote can only confirm them, not reshape them. The academic literature on early supplier involvement (Petersen, Handfield and Ragatz, in the Journal of Product Innovation Management tradition) consistently finds that suppliers engaged during design reduce both development cost and lead time, because expensive late-stage changes and tooling rework simply never happen.
2. The supplier you are stuck with. In an engineered category, the geometry often selects the supplier. A part drawn around one shop's process is, in practice, single-sourced. Bring procurement in early and the supplier base is a choice; bring it in at RFQ and it is a conclusion. McKinsey points out that the majority of a product's value is created by suppliers upstream of the OEM — which means the supplier structure you eventually negotiate against is set in design, not in sourcing.
3. The risk you inherit. Single-source tooling, a fragile tier-2 input, a geographic concentration you cannot see — these are trade-offs at design and facts at RFQ. At the request for quotation they are risks you absorb; at design they are decisions you could still shape.
Why late arrival is the default.
It is not laziness. Procurement is staffed and measured around the sourcing event, and engineering is measured on the release date. The handoff looks clean: design done, now go buy it. Deloitte's 2025 Global CPO Survey still finds siloed ways of working — named by 57 percent of CPOs — among the top barriers to procurement's influence. That wall is exactly what keeps procurement out of the room where the cost is actually set. The function is asked to "get the best price" on a part whose price was largely fixed the day the drawing was released.
What early involvement is — and is not.
This is not procurement approving drawings. It is a working session where a supplier, or a category manager who carries supplier knowledge, sits beside engineering and says: here is a simpler geometry that hits the same function, here is a material that is cheaper and available, here is a tolerance we do not need. The point is not to add a gate. It is to move the leverage upstream, where it still exists.
The honest limit.
Early involvement costs engineering's time, and it needs procurement people who can actually read a drawing. Done badly — as a second sign-off — it slows programs and earns resentment. The research is also clear that supplier integration works best with a deliberate process and the right suppliers, not every supplier on every project. The claim here is narrow: for engineered categories, the late-default quietly forfeits the biggest levers. It does not say "involve procurement in everything."
One question before you accept the RFQ.
Next time a custom-part request lands, ask "who decided what this part costs?" before you ask "what is your best price?" If the answer is "engineering, three months ago," you are not sourcing. You are pricing someone else's decision — and you have already lost the part you came to own.