FOLEY STRATEGIC ADVISORY
WEEKLY BRIEF · ISSUE 1 · WEEK OF SEPTEMBER 21, 2026
Almost everyone is piloting. Almost no one has shipped.
Forty-nine out of a hundred procurement teams have piloted. Four run it at scale. Four moves, in order.
Forty-nine out of a hundred procurement teams have run an artificial intelligence pilot. Four have it running at scale. That gap is not a technology gap. It is an execution gap, and it shows up in the purchase order book long before it shows up in a board deck.
Below are four moves, in the order they have to happen. Each one has a Monday version you can start without a budget request.
49%
have piloted artificial intelligence somewhere in the function
4%
have it running at scale, where it changes a decision
1 · SEE IT
Clean the spend file before you buy anything else
Automated classification maps most purchase and invoice records to a category on its own. Run it with no human validation layer and accuracy lands somewhere around 70 to 85 percent. The slice it misses is not random. That is where the duplicate supplier records, the off-contract buying, and the volume you could have consolidated are sitting.
Do not put a chatbot on the category desk first. Put a clean file under it. If you cannot see the spend, you cannot reprice it, and every number you report after that rests on a category boundary nobody checked.
Monday: Pull three years of purchase orders and invoices. Count the lines that are uncategorized or coded to a general ledger account instead of a category. That number is your starting line.
2 · PRICE IT
Should-cost in hours, not a gut call
Teams running structured spend analysis on one unified data set report roughly 8 percent savings on addressable spend, against the 2 to 3 percent that traditional approaches deliver. The Hackett Group puts the productivity and quality lift from artificial intelligence procurement tools at up to 10 percent on top of that.
A should-cost is not a forecast. It is a negotiating position with the math attached: material, labor, freight, margin, and the total cost of owning the thing, refreshed when the inputs move. Judgment still decides. It just stops deciding alone.
Monday: One named category. One should-cost model. One scenario at plus and minus fifteen percent on the main cost driver. Take it into a live negotiation before you scale it.
3 · WATCH IT
Hear the trouble before the force majeure letter
Validated deployments put the useful warning window at roughly two to four weeks for an operational disruption. Financial distress shows earlier than that: falling operating cash flow, stretched payables, and legal filings drift for months before a supplier misses a delivery. Deloitte found 93 percent of organizations still sit at low maturity on this capability.
Be careful with anyone selling you six months of foresight. Two to four weeks is enough to qualify an alternate, move a buffer, or get on a plane. It is not enough if the signal lands in a quarterly review instead of an operating rhythm.
Monday: Name the ten suppliers whose failure stops a line. Turn on continuous financial and news monitoring for those ten. Not five hundred.
4 · STAFF IT
Get your best buyers off the invoices
Procurement professionals estimate that about ten and a half hours of their week could be automated today. That is more than a full working day, per person. Meanwhile the function trails the rest of the back office on broad rollout: 13 to 18 percent, against 32 percent in customer service.
The goal is not fewer people. It is people on the file instead of in the file. Order matching, invoice processing, contract extraction, and compliance checks should run themselves. Supplier relationships, negotiation strategy, and category strategy should not.
Monday: Ask every buyer what they did last week that a machine could have done. Total the hours. That is your business case, and you already own the data.
One more, and it will not help anybody sell you software
Eighty-three percent of procurement organizations have no enforced policy on what supplier data can be pasted into an artificial intelligence tool. Pricing, competitive bids, contract terms. This is the cheapest item on the list: a policy is a decision, not a budget line. Most pilots that die do not die in the technology review.
What is on fire in your seat?
Spend you cannot see. A supplier clock running down. A should-cost you are still running on instinct.
The question is not whether artificial intelligence changes supply chains. It is whether you lead the change or scramble to catch it.
John Foley
Foley Strategic Advisory
Want the next Brief in your inbox? Subscribe on the Weekly Brief page to get every Monday issue.
Subscribe to the Weekly Brief