There is a conversation that happens in most organizations once a year, usually in the weeks after a sourcing cycle closes, and it goes the same way almost everywhere. Procurement presents the savings it delivered. Finance listens, agrees the negotiation was good work, and then says some version of: I believe you, and I cannot find that money.
Both people are being honest. That is what makes the conversation so hard to resolve, and why it tends to repeat annually rather than get settled. The savings were genuinely negotiated. The budget genuinely did not fall. The gap between those two facts has a name in procurement circles, savings leakage, and it is almost never caused by the thing people reach for first, which is doubting the negotiation.
The negotiation is usually the strongest part of the whole chain. What is weak is everything that happens after the signature, which nobody was assigned to own.
The two numbers are measuring different things
A negotiated saving is a statement about a price. It says: for this good or service, on these terms, the unit we will pay is lower than the unit we were paying. It is measured at the moment of agreement, against a baseline procurement chose, on volumes procurement assumed.
A realized saving is a statement about cash. It says: less money left the organization this period than would have left it otherwise, and the difference is visible somewhere a controller can point to.
Those two claims are related, but a great deal has to hold true for the first to become the second. The volume has to materialize roughly as modelled. The organization has to actually buy through the contract that was negotiated. The specification has to stay put. And when the money does stop leaving, something has to remove it from a budget line, or it simply sits there and gets spent on something else, which is not fraud, or even carelessness. It is what unallocated budget does.
The useful reframe is this: a procurement team can only produce the first number. Producing the second one is an operating problem that runs through finance, the budget holders, and the people doing the actual buying. Treating it as a procurement performance question is why the annual conversation stays stuck. Procurement is being held to an outcome it does not control the second half of.
The five places savings leak
Leakage is rarely one large failure. It is usually several ordinary ones, each individually defensible, compounding across a spend base.
The baseline was never agreed with the people who would have to find the money
This is the most common and the most fixable. Procurement builds a baseline (last year's price, a market reference, the pre-negotiation quote) because it needs one to measure against. Finance is often not in that conversation, and discovers the baseline for the first time in the savings report, at the exact moment it is being asked to accept a conclusion drawn from it.
At that point the baseline is not a shared measurement, it is a claim under review. And there is almost always a reasonable objection: last year's price included a one-off, or the market reference is not how we buy, or the pre-negotiation quote was an opening position nobody ever intended to pay. Every one of those objections may be fair. But they arrive after the work, when relitigating the number costs more than agreeing it would have cost beforehand.
A baseline agreed with finance before the negotiation starts is a slightly worse number that everybody accepts. That trade is worth making every time.
The contract was signed and the buying did not change
A negotiated agreement is permission to buy at a price. It is not a mechanism that causes anyone to do so.
If a category manager somewhere continues to order from the incumbent supplier out of habit, or a regional office keeps its local vendor because the relationship works, or a department buys through a marketplace because the requisition path is faster than the compliant one, the negotiated price applies to a fraction of the spend it was modelled on. The savings report will typically not know this, because it was calculated from projected volume at signature and never reconciled against what was actually purchased.
Off-contract buying is worth understanding rather than just prohibiting. People route around a procurement process for the same reasons they route around any process: it is slow, it does not cover their case, or the compliant option is genuinely worse for the outcome they are accountable for. Each of those is a different problem with a different fix, and none of them is solved by another reminder email. This is where a spend problem turns out to be a process design problem wearing a procurement costume.
Volume assumptions nobody owned
Tiered pricing and committed volumes make the savings arithmetic dependent on a forecast. Someone produced that forecast, usually under time pressure, usually optimistically, and usually without being told that a contract term would be built on it.
The arithmetic is unforgiving. Price a tier against a commitment of 10,000 units and buy 6,500, and the effective unit price can land above the pre-negotiation quote even though the negotiated rate on paper is lower; the saving in the report and the premium in the invoices are both real, measured against different volumes. When the volume lands lower, the discount tier is missed and the effective unit price is worse than modelled. When it lands higher, the organization may be over-committed to a supplier whose leverage just improved. Either way the saving on the page was calculated at a volume that did not happen, and nobody was accountable for the assumption because it was never framed as a commitment; it was framed as an input.
The fix is unglamorous: name the person who owns each volume assumption in the contract, and put a review date on it before the tier resets.
Price came down and total cost went up
The narrowest way to measure a sourcing decision is unit price, and it is the way most savings get reported, because it is the number that is easy to defend.
Total cost includes the things that unit price excludes: a longer lead time that has to be absorbed by inventory, a lower specification that increases rework or failure rates, a supplier that requires more management attention, an implementation or switching cost paid once but paid in full, quality variance that lands somewhere downstream as a warranty or service problem. A saving that creates work elsewhere has not removed cost from the organization. It has moved it into a budget that does not report to procurement, which is precisely why it stops being counted.
The related version of this is risk. Consolidating a category to a single supplier often produces the best price on the page and buys a single point of failure that will not appear in any savings number until the year it does. Consolidation can be exactly the right call, but it should be a decision the organization makes deliberately, with the fragility named, not a side effect of optimizing the metric that happened to be measured.
The saving was real and the budget kept it
This one is different from the others, because nothing went wrong operationally. The negotiation held, the buying complied, the volume landed, the total cost genuinely fell. And the budget line stayed exactly where it was, because no mechanism existed to take it out.
Money that stops being needed for its original purpose does not return itself. It gets spent, reasonably, on the next thing the budget holder needed, and by the time anyone reconciles, the saving is invisible and indistinguishable from having never happened. Finance is right that the money is not there. Procurement is right that it was saved.
Whether savings should be swept out of budgets or reinvested is a legitimate policy choice and organizations land in different places for good reasons. What is not legitimate is not having decided. If no one has answered the question, the default answer is that savings are reinvested silently and reported as if they were banked, and that is the version that erodes trust in every subsequent number procurement produces.
What closing the gap actually requires
The pattern across all five is the same. Each is a handoff between procurement and someone else that nobody wrote down. So the remedies are boring and structural rather than analytical.
Agree the baseline before the negotiation, with finance in the room. Write it down, including what it excludes. This single step removes most of the annual argument, because the argument is nearly always about the baseline rather than the negotiation.
Measure contract compliance, not just contract coverage. Coverage is what share of spend has an agreement behind it. Compliance is what share of spend actually went through one. Only the second predicts whether the saving arrives, and the gap between them is where the diagnosis lives.
Report leakage rather than suppressing it. A savings number that shows negotiated, then compliance, then realized, is less flattering and vastly more useful. It survives contact with finance, it points at which of the five failures is happening, and it makes the following year's number credible.
Decide the budget mechanism in advance. Swept, reinvested, or split: pick one, in writing, before the sourcing cycle rather than after. This is a governance decision, not a finance formality, and it belongs with the same people who own the rest of the control environment.
Give the realized number an owner. Not procurement, whose accountability ends at the agreement. Someone with authority over both the buying behaviour and the budget line, reviewing on a defined cadence, with the authority to escalate when the two numbers diverge.
Underneath all five is a data question, and it is worth naming plainly. Most organizations cannot measure realized savings because their spend data will not support the comparison: categories that are inconsistent across systems, supplier records that do not deduplicate, invoice lines that cannot be traced back to the contract that governs them. That is not a procurement failure either. It is a data strategy problem that happens to surface here first, and no amount of effort in the sourcing cycle will fix a measurement the underlying records cannot produce.
When the leakage is the right answer
Not every gap between negotiated and realized is a defect, and treating it as one produces its own distortions.
Sometimes the specification changed because the business changed, and buying the cheaper thing would have been the wrong decision. Sometimes an organization consciously chooses to reinvest a saving into the same category (better service levels, a more resilient supply base, a supplier relationship worth keeping), and that is a strategy, not a leak. Sometimes the volume did not materialize because demand fell, which is a business outcome, not a sourcing failure.
The distinction is not whether the money arrived. It is whether the divergence was a decision somebody made and can defend, or a surprise discovered a year later. An organization that can explain each gap is in good shape even when the realized number is well under the negotiated one. An organization that cannot explain any of them has a measurement problem that will keep producing the same annual conversation, regardless of how well it negotiates.
Where this fits
Procurement work that stops at the signature is only half-finished, and it is the half that is easiest to do well and hardest to bank. The sourcing strategy, the spend analysis that makes a real baseline possible, and the negotiation approach for the vendors that matter most are procurement consulting; making the buying behaviour actually change afterwards is closer to change management, and the ability to measure any of it credibly usually depends on the state of the spend data underneath. Where a sourcing decision is being made on a number no one can reconstruct, the more useful first question is what the data can actually support rather than what the model assumes.
If your savings reporting and your budget have stopped agreeing with each other, the gap is diagnosable, and it is usually four or five specific handoffs rather than a general failure of rigour. Start a scoping conversation: we would rather tell you the negotiation was sound and the measurement is the problem than sell you a sourcing cycle you have already run well.