Margin does not drop in one event. It leaks through a handful of specific, individually manageable disconnects between what a program was quoted at and what actually happens between quote and SOP. Each one, on its own, looks small enough to absorb. None of them show up on a single dashboard, because no system in a typical supplier's stack tracks actual performance against the original quote. That is why finance usually sees the damage only after actuals post, when the windows to recover it, an engineering change negotiation, a price-down conversation, a PPAP milestone, have already closed.
Global automotive suppliers are operating on thin margins to begin with. Roland Berger's 2025 Global Automotive Supplier Study put the average EBIT margin at just 4.7% for 2024, which means a leak of one or two points can consume most of a program's profit.
Two high-impact, hard-to-see leaks are stale raw material cost assumptions and unplanned premium freight, both of which accumulate in small increments that no single system is built to total against the original quote.
Siemens' 2024 True Cost of Downtime report puts unplanned downtime in automotive manufacturing at roughly $2.3 million an hour, more than $38,000 a minute, which is why suppliers will absorb enormous premium freight costs rather than risk a shutdown, often without anyone tracking the cumulative hit to program profit.
ERP, PLM, and quoting or CRM systems each do their own job well. ERP tracks actual costs, PLM (where it exists) tracks engineering data, purchasing and quality systems track their own transactions. None of them are built to hold "what we quoted" as a living reference point.
Closing the gap takes a named owner and scheduled reconciliation checkpoints, comparing quote and actuals at defined points across a program's life, not just a better individual system.
Early in my career, I worked in IT for a mid-sized Tier 1/2 plastics supplier, building and supporting the systems the business ran on. I was not on the finance or program side, but I sat in enough meetings and heard enough conversations between plant managers and the owner to see the pattern up close.
Every week, one plant's team walked into a meeting with the owner and presented numbers off a manually built Excel file the team called the "cost tracker." It broke costs down to a level of detail finer than anything the ERP reported, and the team treated it like their secret weapon. It was also entirely manual, entirely dependent on someone choosing to enter the bad news, and prone to every problem a spreadsheet-based process has.
The owner had two phrases he used constantly. One was "don't be a potted plant." If you saw a pattern forming, like premium freight going out three or four weeks in a row, you were expected to say something immediately, not just watch it happen. The other was that the company had too many "one-armed sweaters," systems and processes that were only ever half built, so they could only ever do half the job they were meant to do. Both phrases turned out to describe the same underlying problem from two different angles: a technical one and a human one.
Plastic injection molders quote piece price based on a resin cost assumption at a point in time. Resin prices, like most commodities, move. When virgin resin costs climbed faster than the quote assumed, one of the ways suppliers offset the gap was regrind: reprocessing the sprues, runners, and scrap generated during molding, and re-blending that material back into the feedstock at an engineering-approved ratio, to reduce how much new virgin resin a part actually required.
Regrind helps, but it has limits. Not every part can tolerate a high regrind ratio without risking spec or cosmetic issues, and there is a ceiling to how much of a resin price spike it can absorb. When the price move outran what regrind could offset, and the supply contract either lacked an economic price adjustment clause or the pass-through was never actually operationalized, the difference came straight out of the program's margin, dollar for dollar against the quoted material cost, every month the price stayed elevated.
The ERP we ran, a commercial system on an IBM mainframe, tracked actual material purchases and cost variance every month. That was its job, and it did it well. But that variance was measured against standard cost, an internal accounting number, not against what the customer had actually been quoted. The ERP had no concept of the original quoted assumption for a specific program; that number lived in a quoting file, created once, and never wired back in as something to compare against. So a material cost swing could look clean in the ERP's own numbers while still quietly eating into a program's quoted margin, and nothing in the system would flag it. That is exactly what the cost tracker was built to do. It carried a P&L in three columns side by side, quote, an internal working number (target), and actual, updated at every review, so someone could actually see the gap the ERP couldn't.
The second pattern was harder to see coming and, by the account I heard, did more damage. When a production issue caused a delay, the plant faced a choice: risk shutting down the OEM's line, or pay to expedite shipments, air freight, dedicated trucks, whatever it took, to keep parts arriving on time. Siemens' 2024 True Cost of Downtime report, based on interviews with maintenance, engineering, and IT professionals at large industrial companies, put the cost of unplanned downtime in automotive manufacturing at roughly $2.3 million an hour, more than $38,000 a minute, having doubled since 2019. Against that, premium freight is usually the cheaper option in the moment. It is a rational decision every single time you make it.
The problem is what happens when you make that decision every week for months. Each expedited shipment gets recorded as a freight cost, same bucket as routine freight, on the same invoice cycle as everything else. No system totals those expedites against the specific program's quoted margin. So the cost accumulates, week after week, until it has quietly consumed the entire program's profit, and in the cases I remember hearing about, it stayed that way for a long time. The way it usually surfaced was not through an internal report. It was a customer threatening to pull business at contract renewal, or a new plant manager taking over and finding it during a routine review.
We had no PLM system at the time either of these stories took place. Engineering data lived in whatever CAD platform the OEM used, most often Catia, entirely outside our own systems. We had a homegrown purchasing system for approvals and a homegrown 8D system for quality issues. Every one of these systems did its own narrow job competently. None of them was built to answer the question that actually determines program profitability: does what is happening right now still match what we quoted?
Even the data that did exist was not live. We made an attempt to automate the cost tracker with data from the ERP. The ERP built its cost and bill-of-material data overnight, and IT imported it into the reporting side once each morning, Monday through Friday. That is a daily snapshot, not a running total, and nothing in that pipeline was designed to notice that this week's number looked like last week's, and the week before that. Only a person scanning several weeks side by side would catch a pattern forming. That is exactly why the culture piece mattered as much as the technology piece. The system could show you a number. Only a person willing to compare this week's number to four weeks ago, and say something, could catch a trend before it became a program-ending one.
That is the structural version of "one-armed sweater." A sweater is built to keep both arms warm. A system built to track actual costs, or engineering changes, or quality corrective actions, does exactly what it was designed to do and nothing more. The reconciliation against the original quote was never any single system's job, so it fell to a manual workaround, and a manual workaround only works if the people running it are willing to surface bad news before it compounds. That is where "don't be a potted plant" comes in. The technical gap and the human gap were the same gap. People often did not flag the freight pattern early because doing so meant admitting a problem in front of the owner, so they tried to absorb it or make it up elsewhere, and by the time it could not be hidden anymore, the recovery window on that program was long closed. This is not a problem that ended with better tools either. The same structural gap exists in modern ERP, PLM, and CRM stacks. They are individually far more capable than what we ran, but they still do not share a common, living definition of "the quote" to reconcile against, which is why this leak still shows up in supplier P&Ls today.
Buying a bigger or newer system does not close this gap by itself. Adding more transactional systems without something that explicitly bridges them back to the original quote can multiply blind spots instead of closing them.
What actually works is treating the quote as a persistent object with a named owner, usually a program finance lead, who is responsible for reconciling actual cost performance against it at defined checkpoints across the program, reviewed on a regular cadence, before a program's actuals have had time to drift far from what was promised.
The fix is procedural before it is technical. The technology only helps once someone has decided that reconciliation is a job, not a hope.
Q: What is the main reason margin leaks between quote and SOP go undetected?
A: ERP, PLM, and CRM or quoting systems each track their own transactions accurately, but none of them are built to hold the original quote as a live reference point, so nothing in the stack flags a widening gap until someone builds a report specifically to look for it.
Q: How much can premium freight actually cost a supplier during a production delay?
A: Siemens' 2024 True Cost of Downtime report puts unplanned downtime in automotive manufacturing at roughly $2.3 million an hour, more than $38,000 a minute, which is why suppliers will absorb significant expedite costs rather than risk a shutdown, often without tracking the cumulative hit to a program's margin.
Q: Can rising raw material costs be passed through to OEM customers?
A: Some supply agreements include economic price adjustment or index clauses that allow for pass-through, but not every supplier has negotiated that protection, and even where a clause exists, someone still has to track the variance and invoice for it, which is a process gap as much as a contract one.
Q: Does regrind fully offset rising resin costs for plastics suppliers?
A: No. Regrind can meaningfully reduce virgin resin consumption, but not every part can tolerate a high regrind ratio without risking spec or cosmetic issues, and there is a ceiling to how much of a price spike it can absorb before the rest hits margin directly.
Q: What is the first practical step to closing the quote-to-actual gap?
A: Assign a named owner responsible for reconciling actual program costs against the original quote on a regular cadence, tracking quote and actual side by side, rather than relying on a one-off manual report or waiting for a customer or a new plant manager to surface the problem.