What actually changed
Buying software made sense when building it was expensive. That was the entire argument, and it was a good one for thirty years.
Building is not expensive anymore. I built twenty-two financial tools in a weekend — not as a stunt, but because owners kept flying blind while the software that would have helped cost more than the problem. The vendors pricing off the old assumption have not repriced.
There was always a second problem too: you buy a platform designed for a median company that does not exist. You pay for a hundred features, use six, and cannot change any of them.
The stack, in order
Each layer depends on the one beneath it, and buying out of order is the most expensive mistake here. The verdicts below are where I'd land today — not where I'd have landed in 2022.
The cost everyone gets wrong
Here is where the “just build it” crowd is as wrong as the vendors. Creating software got cheap. Owning it did not.
Someone has to keep it running, patch it, and understand it at eleven at night during close when it breaks. That cost is real, recurring, and almost never in the spreadsheet people use to justify building.
So the calculus did not disappear — the variable that dominates it moved. It used to be build cost. Now it's ownership cost. Buy the layers where somebody else carrying that burden is worth the money: the regulated ones, the audited ones, the ones touching your bank.
Four mistakes that cost the most
Buying the dashboard before the close is clean
A reporting tool pointed at unreliable books produces fast, confident, wrong answers — and makes the problem harder to see, because the output looks professional. Clean the close first. Every time.
Buying a subscription for something you could commission in a week
This is the new one, and it is now the most common. Ask what a tool actually does. If the answer is 'stores our data and shows it back to us in a particular shape,' you are renting a shape. That was a reasonable trade when building was expensive.
Buying an ERP to fix a process problem
If the current process is undefined, an ERP encodes the confusion at considerable expense. Systems make an existing process faster and more rigid — they do not invent one.
Treating a spreadsheet as a failure state
A weekly-updated thirteen-week cash forecast in a spreadsheet beats an unmaintained forecast in an expensive tool, every time. The question is whether the artifact exists and is current, not what it was built in.
The skill that just became scarce
If buying is no longer the decision, then picking a vendor is no longer the skill. What matters now is specifying — knowing exactly what you need, in what order, and being able to describe it precisely enough to have it built.
That is an operator's skill, not an engineer's. It comes from knowing how the money actually moves through your business. Which is the same thing that made a good CFO valuable before any of this — it just pays differently now.
Start by finding out what's actually broken
Before buying or building anything, it's worth knowing which part of your finances is actually costing you. The diagnostic walks your numbers and tells you what to fix first — and it is usually not software.