In most finance calendars, the indirect savings target for next year is set weeks before anyone has a defensible baseline to measure it against.
The figure gets built during the planning cycle, when visibility into the spend base is at its thinnest, and it gets delivered in the first quarter, when there is the least runway to work a category properly. By the time someone can say what the run rate actually is by category and by site, the number it has to beat is already in the plan.
This is not a failure of planning discipline. The budget has to close, and no one is going to hold it open while procurement reads the ledger. But it does shape the first quarter, which frequently gets spent reconciling a committed figure against a spend base nobody has normalized yet.
A few things make that reconciliation less painful.
“Nothing left to cut” is usually a statement about visibility
The standard response to a fresh target is a reasonable one. We already ran a cost-cutting push, and there is not much left to find.
That is frequently true of everything the last exercise could see. Headcount, T&E, software seats, the top contracts by annualized value. Those get worked first because they are easy to see in the ledger and the savings are straightforward to book.
What survives a cost-cutting push is generally not what got defended. It is the tail: the categories that are individually immaterial and collectively significant, and that require category depth rather than leverage. Freight accessorials and surcharge schedules. Facility services on evergreen terms. Small parcel. Packaging. Indirect materials consumed at the site level and expensed without a PO. Each one rounds to nothing against revenue. Added up, the tail is rarely small, and it is rarely where the last push was aimed.
The useful question at this point in the cycle is not whether there is anything left. It is which parts of the spend base the last exercise never had a clean view of.
Normalize the ledger before sourcing anything
A target is only measurable against a baseline that has been normalized for volume and mix. In multi-site organizations, that work is usually heavier than expected, and it is the gating item for everything downstream.
Part of the difficulty is structural, and the pattern repeats across multi-site companies. Approval thresholds are documented but not enforced in the purchasing workflow. Buying follows habit rather than the org chart, and the AP ledger reflects that rather than the reporting lines.
That matters for how a savings number converts. Where sites buy the same category independently, a negotiated rate is a contracted rate, not a realized one. Realization depends on order flow moving to the agreement, which is an adoption problem rather than a sourcing problem, and it tends to be where the variance between forecast and booked savings opens up.
In practice, the highest-yield exercise at this point in the year is frequently not a sourcing event. It is a vendor-by-site cut of twelve months of AP, deduplicated for vendor naming, with intercompany and pass-through stripped out.
Treat the January increase notices as benchmarking input
Supplier increase letters arrive in Q4 against a January effective date, and they are generally read as a cost event rather than as a data set. They are one of the few unsolicited signals a company gets about where its own cost base sits relative to the market.
For context on the current environment, the Institute for Supply Management put its manufacturing prices index at 75.2 in October 2026, down from 77.9 in September, with the headline PMI at 55.5. Input costs are still rising across most of the manufacturing base, just somewhat less quickly than they were.
The relevance is not predictive. It is calibration. In a broadly inflationary input environment, an increase that tracks the index is a market condition, and an increase materially above it is a negotiating position. Separating the two is most of the analytical value, and it requires an agreed external benchmark rather than a view.
This is also the point where savings methodology gets tested. Where a category is index-linked, a year-over-year price decline that lags the index is not performance, and booking it as such will not survive an audit. Variance to an agreed market benchmark, fixed before the work begins, is the defensible construction. Variance to last year’s invoice is not.
Compare prices across sites, not just over time
Most purchase price variance reporting compares this year to last year. Comparing sites to each other is often more productive and takes considerably less time: select a handful of high-frequency SKUs and compare unit price paid by location over the trailing twelve months.
Same part number. Frequently the same distributor. Materially different unit economics.
The causes are mundane. Agreements executed at different points in the cycle, volume tiers reached at some sites and not others, an acquired location still transacting on its legacy terms, account coverage that changed hands. None of it implies a control failure. It persists because no individual site can see the comparison, and at the consolidated level the category resolves to a single line that looks unremarkable.
Two things recommend this cut during budget season. It yields a quantified, auditable number rather than a modeled one. And it tends to resolve the question of whether the tail has in fact been worked.
Protect the baseline, not just the number
Most of the credibility risk in a savings program sits in the baseline rather than in the sourcing. Targets met through deferral, scope reduction, or volume decline will reconcile cleanly in the plan and fail to show up in the run rate, generally surfacing a cycle later.
The practical guardrails are familiar: fix the baseline period and the volume normalization before the work starts, agree in advance which categories are index-linked and against which published series, report realized savings and cost avoidance separately rather than in the footnotes, and reconcile booked savings to the general ledger on a defined cadence instead of at year end.
None of that is novel. It is just easier to agree in November than to litigate in June.
What the rest of Q4 is actually good for
Not sourcing. A competitive process initiated in December rarely closes well.
It is good for three low-cost items that compress Q1. Build the normalized twelve month spend file by vendor and by site. Produce a renewal calendar for indirect agreements with notice windows and auto-renewal terms flagged, since a missed notice date removes a category from scope for a full term. And triage the increase notices already received into those tracking the relevant index and those exceeding it.
None of this commits anyone to a decision. All of it means that when the committed figure has to become a category plan in January, the baseline is already defensible.
The target is going to be set before the baseline exists. That is a function of the calendar. What is controllable is how much of the baseline is built by the time the work starts.
Source: Institute for Supply Management, Manufacturing PMI Report on Business, October 2026 (released November 2, 2026).
