FIELD NOTE |
November 18, 2025
The Cheapest Option Can Be an Expensive System
A purchase can be reasonable on its own and still contribute to a system nobody would have chosen deliberately.
Procurement decisions are often compared in columns: license cost, implementation fee, discount, contract term. Those numbers matter. They're also unusually easy to see.
The other costs tend to arrive somewhere else.
Integration work. Duplicate data. Training. Manual reconciliation. Support. Security review. Process changes. Administration. Reporting. Renewal negotiation. Eventually, the cost of leaving.
The purchase price may belong to one budget while the cost of making the purchase actually work is distributed across everyone else.
I've seen tools justified because they reduced a software line item while quietly requiring people to become the integration layer.
A person exports the data. Another cleans it. Someone reconciles the definitions. A manager explains why two dashboards disagree.
None of those hours appear on the vendor invoice. The software is still cheap.
But the people have become middleware.
This kind of cost accumulates.
Another field. Another integration. Another workflow. Another exception. Another source of truth that is almost the same as the previous one.
Total cost includes the exit
I also want to know what happens when the relationship ends. Can we export our data cleanly? Who owns the configuration? How much of our process has been shaped around proprietary behavior? What knowledge exists only with the vendor?
A low entry price paired with a punishing exit is not a cheap system. It is deferred cost.
Some costs aren't things we will pay. They're choices we will have less freedom to make later.
The more deeply a tool becomes embedded in the way an organization works, the more expensive it may become to choose differently.
The purchase price tells you what it costs to say yes. It tells you very little about what that yes will require afterward.
None of this means fewer tools are automatically better.
A simpler technology stack can be easier to secure, maintain, understand, and connect. But simplification has costs too. Sometimes another tool creates a capability worth the complexity it introduces.
The question is what the tool requires from the surrounding system—and whether what it makes possible is worth that requirement.
Paying more doesn't solve this either. An expensive platform is waste if the organization can't use the capability it was bought to provide.
Cost isn't evidence of value in either direction.
The expensive part is often not the tool itself. It's the relationships the tool creates with everything around it.
Finance reconciles it. Security reviews it. Operations administers it. Data engineering moves information in and out. Managers explain discrepancies. Users invent workarounds.
Each of those costs may be entirely reasonable on its own.
The combined system may not be.
That's the problem with evaluating purchases one at a time. An organization can make a series of individually defensible decisions and still end up with an environment nobody would have chosen deliberately.
Nobody sets out to make people the integration layer. Nobody deliberately chooses a technology portfolio that requires more and more of the organization just to keep it coherent.
It happens one reasonable decision at a time.
So when I'm evaluating a purchase, I want to know what it costs.
I also want to know what the rest of the system will have to become in order to live with it.