What is the difference between procurement savings and cost avoidance?

Saving

The organisation pays less than it did, measured against an agreed baseline, for a requirement that has not materially changed.

Cost avoidance

A cost that would have arisen does not, or an increase is reduced or removed. The organisation is better off than it would have been and the budget stays where it is.

Cash-releasing

Whether the money can physically be taken out of a budget line, in a stated month, against a stated cost centre. A benefit can pass the savings test and still not be cash-releasing: released staff time in a service whose establishment will not reduce, a lower notional recharge, a reduced accrual. Public sector finance separates cash-releasing from non-cash-releasing before it separates savings from avoidance.

Recurrent

Whether it reduces next year's run rate. A rebate, a one-off credit or the settlement of a disputed charge is real and is not recurrent. Reporting one as though it were overstates the position by a year and it is discovered exactly a year later.

The confusion is rarely about the concepts. It is about which one a particular event is, and that argument is nearly always about the baseline.

One question settles most disputes. Can the budget holder give up the money?

• If yes, it is a saving. The spend was going to happen at the old price and it will now happen at a lower one.
• If no, it is avoidance. The money was never in the budget, or the old figure still stands and still has to be funded.

If nobody has to give up any budget, nothing has been saved. Something has been prevented, which is different and still worth reporting.

Applied honestly the test reclassifies a good deal of what is normally reported as savings, and it does so in the direction nobody enjoys. The reason to do it anyway is that the alternative is worse. A finance team told that procurement has delivered a large saving may reasonably remove it from budgets. Where most of it was avoidance the money was never there, the service takes the cut, and the next savings number procurement reports is read with that in mind.

In the public sector the test needs one adjustment. A budget can almost always be reduced on paper; the question is whether the reduction is deliverable, whether it recurs, and whether the service can still be provided. A cut that is agreed and then quietly funded from a reserve is not a saving, whatever the classification says.

Six situations that come up constantly, and how each should be treated.

1. 01A re-tender comes in below the current contracted priceA saving, for the same requirement at the same volumes. Baseline is the current contracted price at current volumes. If volumes are forecast to rise, say so on the same line, because total spend may go up while the unit price falls.
2. 02A supplier requests an increase and settles lowerAvoidance of the difference, and a cost pressure of whatever was agreed. Both belong on the report. Record the supplier's written request as the evidence, because without it the number is an assertion. Where the increase arrived as a portal notification or a new price list with no covering letter, keep that instead, and where it arrived verbally, ask for it in writing before conceding anything.
3. 03The increase was contractually owedNeither. If the contract contains an indexation clause and the supplier has applied it correctly, they are entitled to it and negotiating it down is a variation, not avoidance. Claiming avoidance against a contractual entitlement is the classification error most likely to be caught by a contract manager reading the report.
4. 04A new requirement, bought for the first timeUsually neither, because there is no baseline. If it has to be reported, the defensible baselines are the pre-tender estimate or the approved business case figure, not the highest compliant bid, which is a number the supplier chose. State which was used.
5. 05The specification is reduced and the price fallsA saving, provided the requirement is genuinely still met. Where the service level has been reduced, record that alongside the number. A saving with a declared service reduction is a legitimate decision; an undeclared one is an incident waiting to happen.
6. 06A regulator raises the specification and the cost goes upNeither, and it is regularly misreported as avoidance. The cost is not optional and there is no version of the decision in which it is not incurred, so there is nothing to avoid. What procurement contributes is best value inside a fixed specification, delivery inside the compliance deadline, and a contract written so that the next change to the standard is a variation rather than a re-procurement. Report it as a cost pressure with the requirement named, and if the market testing produced a lower price than the first quotation, that difference is the number, against a stated baseline.

Avoidance in a period of supplier-led price increases can exceed savings, and a function judged only on cash in such a period looks inactive while doing its most valuable work. The rules that keep both credible are short.

• Two lines, never one. A combined figure invites the question of what is in it, and the answer is always worse than reporting them separately would have been.
• State the baseline on every line, including avoidance, where the baseline is the increase that was requested or the price that would otherwise have applied.
• State the period. Annual benefit, and contract term value separately if it is relevant, never the term value presented as this year's number.
• Flag cash-releasing and recurrent on every line. These are the two things finance reads first and the two most often missing.
• Hold avoidance to the same evidential standard as savings. It is where fabricated numbers live, because nobody validates it, and the reason nobody validates it is that nobody is going to act on it.
• Agree the attribution rule in advance. Where the budget holder made the demand decision and procurement ran the process, someone has to have decided in advance who claims it, or the argument happens in the meeting.

The classification argument recurs, and it recurs at the same point every year: the roll-forward.

An increase that was avoided this year becomes part of the contracted price next year. The avoidance does not repeat, and if the baseline rolls forward to the new price then the same negotiation next year starts from a lower figure and produces a smaller number. That is correct and it looks like decline.

Two things prevent the argument. Agree in advance that the baseline rolls forward to the current contracted price rather than being held at a historic one, so nobody is claiming the same benefit twice. And expect the avoidance line to be lumpy year to year, because it tracks what suppliers asked for rather than anything procurement controls.

Where a saving is genuinely recurrent it should stop being reported after the first year and simply be in the base. A function still reporting a benefit delivered three years ago is padding, and it is the padding most easily spotted by anyone who has the previous reports.