Art of Procurement BTW EP 28: Develop Expense-Specific Systems: Why One Dashboard Can't Manage Every Category
Procurement talks about "the data" as if it's neutral.
It rarely is.
For years, we have talked about "the data" as if it were a single, uniform thing… a stack of invoices, a dashboard of KPIs, a quarterly business review deck handed over by a supplier.
Here's the problem: invoices are curated. Reports are crafted. And, most of the time, suppliers decide what you see… unless you know what to ask for.
In this episode of Buy: The Way…To Purposeful Procurement, Brian Gamble, COO at FineTune and a 30-year veteran of indirect services, joins podcast co-hosts Philip Ideson and Rich Ham to unpack BuyLaw #6: "develop expense-specific systems."
The directive is fairly simple on its surface, but it's also disruptive: no single data set or measurement system works across diverse categories. Uniforms are not utilities. Security is not pest control. Waste is not janitorial supplies. And trying to manage them all with the same playbook guarantees procurement will create blind spots.
Brian has seen those blind spots from both sides up close, first as a regional VP for a national uniform provider, now as an advisor helping clients defend their P&L against quiet leakage. He doesn't mince words: if your definition of "the data" is whatever appears on an invoice PDF, you are operating inside a commercial narrative written by your supplier.
The episode walks through examples that sound almost unbelievable until you realize how common they are. Security "dark hours" where posts go unfilled but still get billed. Pest control programs charging for weekly service where there's been no activity in months. Uniform inventory definitions that vary between suppliers, creating a scenario where 17 cents can be far more expensive than 21 cents, depending on what number you're multiplying.
None of that shows up cleanly on a summary invoice. Which brings us to AI…
As procurement leans more heavily on AI for benchmarking and research, the technology can generate polished, authoritative answers, even when the underlying data is thin or incomplete. But, the quality of the output rises or falls with the quality of the inputs. For example, Brian shares a live demonstration his team conducted internally: a generalist asking AI for "a good price" in a complex service category gets laughable, contradictory answers. Garbage in, garbage out, so to speak. A more informed user does slightly better. When a true category expert feeds AI high-quality, relevant, structured data does the output become meaningfully useful, and even then, it still requires human judgment to separate signal from noise.
This episode also challenges another sacred cow in procurement: not all dollars are created equal. A $100 million utilities category might require minimal management. A $1 million uniform program might require 50 times the oversight. Yet procurement teams are often sized and measured purely by spend under management, not complexity, risk, or management intensity.
If procurement is going to be measured by what actually hits the P&L (as the earlier BuyLaws argue) then they must design contracts, data rights, and reporting structures that allow real validation.
The future of procurement won't be won by those who have the most data. It will be won by those who know which data matters and, perhaps most importantly, why.
Links: