Every large organization I've worked with makes its most consequential commercial decisions through proxies, mostly without noticing.

A proxy is the stand-in you reach for when the real thing is harder to see: the purchase order instead of the commitment, the approval threshold instead of the risk, the category code instead of what's actually being bought, the vendor master row instead of the relationship, the savings number instead of the value created. Every one of them is useful, and every one of them is a container for context rather than the context itself. The trouble starts when the organization forgets which of the two it's holding.

The rule that measured the wrong thing

The 100ml liquid rule at airport security was meant to reduce danger, and it went about that by measuring the size of the container rather than the chemistry of what was inside. Once the substitution was made, the system could enforce itself with perfect consistency while drifting further from the outcome it actually cared about.

That's what proxy logic does. It hands you something cheap, countable and easy to check, so the organization gets very good at optimizing it, and the gap between the stand-in and reality goes mostly unmeasured because measuring the real thing is slower, messier and more expensive. The same pattern runs right through commercial operations.

A purchase order is not the commitment

A purchase order feels definitive because it's visible. It has a number, a date, an owner and a workflow around it, so it looks like the moment spend became real.

The commitment usually happened earlier, in the email thread where the scope got agreed, or the renewal nobody challenged in time, or the operational dependency that made switching vendors unrealistic long before anyone raised a PO. By the time the purchase order appears, the leverage has generally gone. Treat the PO as the start of the story and you miss the months in which the outcome was decided.

An approval threshold is not risk

Thresholds exist for good reasons. They reduce cognitive load, create consistency and help teams move, which is what a decision tool is for. What a threshold can't do is describe reality. A low-value contract can carry serious risk when it touches critical operations, sensitive data or a supplier who'd be hard to replace, and a high-value contract can be fairly benign when the service is standardized and the alternatives are everywhere.

The threshold ends up standing in for the risk anyway, so effort goes into escalating what's expensive and rather less into examining what's consequential. The number is the legible part, and the exposure stays buried in context nobody had time to read.

A category code is not what a thing is

Classification systems are necessary, and finance, procurement and reporting all depend on them. Anyone who has worked inside a large organization also knows how much falls away the moment a real purchase gets flattened into a label.

A category code tells you how a transaction was recorded. It won't reliably tell you what's being bought, what outcome that spend supports, whether similar services sit elsewhere under different labels, or whether one supplier is performing the same function in three business units. That's where fragmentation compounds, because the same vendor shows up as software, consulting, maintenance or managed services depending on who bought it, how they coded it and which system caught it. Optimize from the code alone and the reporting comes out clean while the picture stays incomplete, which makes the organization look more orderly than it is.

A vendor master row is not a relationship

Most systems hold the supplier as a record: one name, one ID, one status, a set of payment details. That's necessary and nowhere near sufficient, because the relationship is a living set of obligations, dependencies, workarounds, performance issues, renewal terms, service credits, price movements and side agreements. Some of it sits in the contract. Much more of it sits in inboxes, meeting threads and the heads of people who may not work there anymore.

So an organization can have a supplier properly recorded and still have no practical visibility into the relationship. They know the vendor exists without knowing what's been promised, what has drifted, what could be challenged, or what's about to renew. The record survives and the reasoning disappears.

A savings number is not value created

Savings is the most useful number in commercial management and the most misleading. Used well, it gives teams a common language for impact. Used badly, it produces a theater of quantified wins that may or may not have improved the business, because a lower price isn't always a better outcome, delayed cost isn't removed cost, and negotiated savings that never reach the P&L aren't commercial value captured. Value also covers the things the metric can't count: avoiding lock-in, protecting service continuity, recovering missed rebates, enforcing credits, and acting before a cost hardens into fact.

Which is how an organization ends up disciplined about reporting savings while staying surprisingly blind to whether its commercial position is improving at all. The metric gets managed carefully while the broader outcome gets assumed.

Why proxies win inside large organizations

None of this happens because people are careless. It happens because proxies are operationally convenient: they fit into workflows, they can be audited and given an owner, they make a dashboard possible, and they reduce ambiguity in an environment where ambiguity is expensive.

The trouble starts when an approximation gets treated as the truth. Teams then optimize the stand-in with genuine effort, following the process correctly, completing the fields, hitting the thresholds, filing the documents and reporting the savings, while the context that would explain what's actually going on stays scattered across email, ERP systems, finance tools, SharePoint, contract repositories and human memory.

What better looks like

Recovering that compressed context is the whole job, and it doesn't start by asking people to write better codes or follow stricter workflows.

In practice it means connecting the signals a business already produces across its vendor network: spend, contracts, communications, obligations, performance history, renewal timing and relationship activity. It means holding why a choice was made alongside what got filed, what has changed since, and where the real leverage still sits.

That's the shift we care about at Cotiss. We're building Commercial Intelligence so an organization can see the full shape of its vendor network instead of the official snapshots left behind in systems of record. Once the context is connected, the proxy goes back to its proper role as a helpful clue rather than the decision.

The question worth asking

If you run finance, procurement or operations, the question is where in your own organization you're managing the proxy as though it were the thing itself.

It gets uncomfortable quickly, and useful about as quickly, because once the pattern is visible you start seeing it everywhere: how often the visible artifact arrives after the real decision, how often a clean report depends on a messy reality being left out, how much commercial value goes missing through systems that preserve the container and discard the chemistry.

What I still can't size is how much of a large business runs on stand-ins nobody has gone back to check. That's the part I keep pulling at.

The proxy problem: why large organizations keep managing the container, not the chemistry