The saving is booked. The cost isn't.

Deloitte has published another report on the hidden cost of poor supplier performance, and my feed treats it like a discovery. 3M has spent more than one earnings call explaining what quality does to a P&L once it stops being absorbable. Neither is news to anyone who has run a plant: the headline figure gets paid out monthly, in the premium freight invoices, the sorting cell at goods-in and the four weeks of buffer stock held against a supplier who cannot reliably hit a Tuesday. That cost is not hidden. It is unbooked.

I inherited a textbook case at a plant I ran quality for. A resourcing trophy: a bracket, four million pieces a year, piece price down €0.11. Purchasing booked €440,000 of savings and took the recognition. Within eighteen months the true landed cost of that part ran six figures above the previous sourcing – sorting at goods-in, two air-freight recoveries, a 41-minute line stop. Purchasing got the plaque; the plant got the sorting cell.

Where the cost actually sits

Supplier failure cost in a typical plant lands in five buckets. Almost none of it is ever booked against the supplier who caused it:

• Premium freight – the expedited truck that disappears into the logistics variance.
• Sorting and rework at goods-in – operators pulled off value-added work, absorbed as plant overhead.
• Buffer stock and line-stop minutes – capital parked against unreliability. The minutes nobody prices.
• Engineering hours policing 8Ds – your SQEs rewriting a supplier's fifth draft because the 8D describes containment, not cause.
• Shared warranty carry – one pot, rarely attributed by source, so the expensive supplier hides inside the average.

Then there is PPM, the metric that flatters the expensive supplier. It counts defective parts per million, weighted by nothing that carries a € sign. A supplier at 300 PPM on cosmetic faults caught at goods-in looks worse than a supplier at 60 PPM whose rare defect is functional and stops a line for half a shift. Frequency is measured. Consequence is not.

You negotiate the price once a year. The cost arrives every day – it just never carries an invoice number.

Two scorecards, no owner

Purchasing is graded on piece-price delta. Quality is graded on PPM. Both numbers are real; neither is total cost. The gap between them belongs to nobody, and in most plants that gap is exactly where the money goes.

When I built the supplier quality system for the SNOP greenfield plant – 900+ people, quality function from zero – the single most productive thing we did was book failure cost per supplier. Every bucket above, per supplier, per month, on one page. The first page was educational: the cheapest supplier on the purchase order was the costliest line in the plant. Once that was visible, the 70% defect-cost reduction we delivered was less a matter of negotiation than of arithmetic. You point supplier development where the euros are – and the euros had been pointing at a supplier nobody worried about, because the price was good.

The mechanism that makes it stick is one COPQ figure per supplier, jointly signed by the purchasing director and the quality director. The joint signature is the whole trick. A number owned by one function gets argued with; a number owned by two gets acted on. It also forces the honest trade-off every supply base faces: you can genuinely develop two suppliers, not twenty. The total-cost ranking tells you which two.

Audit the number, not just the part

VDA 6.3 tells you whether a supplier's process can hold the tolerance. A total-cost review tells you whether the relationship is worth keeping. Run both, annually, as a pair. The audit without the cost review keeps suppliers who are capable and ruinous; the cost review without the audit cuts suppliers who were merely fixable.

At Witte Automotive the supplier work ran on QRQC and A3, and the lesson was counterintuitive: the fix was not policing more 8Ds. It was policing fewer – by making the cost of failure visible to the supplier's own management. When their general manager saw, in €, what his quality department was generating inside our plant, the 8Ds thinned out, because his questions started arriving before ours did.

One caution from de-escalation work. When this figure first surfaces in your organisation, do not walk into a supplier meeting with a slide titled "your cost to us" and a tribunal tone. You will win the meeting and spend the next year not being told things. Early warnings dry up, and you start learning about problems from your own customer instead of your supplier. Present the number as a shared ledger, not a prosecution exhibit. The objective is a supplier who calls you before the truck leaves, not after the line stops.

On consulting engagements I run through FOREAST, the first document I ask a client for is not the PPM report. It is the freight invoices.

The discipline test

Here is the test, and it takes ten seconds. Name your three most expensive suppliers by total cost. Today, from memory. If you cannot, someone else is naming them for you on your P&L – in the freight variance, the sorting cell, the buffer stock, the warranty pot. At plant scale the unbooked cost shows up as premium freight and stopped lines; at corporate scale it ends up on an earnings call, in front of analysts, wearing a number with a comma structure nobody budgeted for. The Deloitte report is a mirror, not news. Every plant already runs this experiment monthly. The only question is whether anyone writes the results down.