Do you need a FinOps tool? Probably not yet

Most growing companies do not need to buy a FinOps tool first. Your cloud provider's cost reports show where the money goes; the return comes from treating that spend as capital, with a named owner for each cost. Buy a platform first if you use several clouds or AI model providers.

The question usually arrives as a purchase decision. Cloud spend has grown faster than revenue for two quarters, someone on the leadership team has seen a demo, and the choice on the table is which cost platform to buy. It is a reasonable instinct. It also treats a visibility problem and a behaviour problem as the same problem, and only one of them comes in a box.

For most growing companies, the answer is not yet. A FinOps tool can show where cloud money goes and carry out actions you have approved in advance, such as switching off idle test systems overnight. It cannot decide who owns each cost or who is allowed to change it, and most of the return depends on those two decisions. If you treat the bill as capital you are investing, the conversation turns to which spend should grow and which should stop.

What a Tool Gives You

FinOps tools fall into a handful of types, and most companies already have the first:

Your cloud provider's own cost reports show spend by account and by tag (the labels engineers attach to each resource), forecast the month and send budget alerts.
Cost visibility platforms pull several providers into one view and share out common costs by an agreed rule.
Commitment tools estimate how much predictable usage you can safely lock in for a discount. Depending on the product, they recommend the purchase, manage it or make it for you.
Container cost tools split one shared pool of servers between the teams that use it.

Commitment tools lower the price of what you run, and they are worth having once your usage is steady, though a commitment is still paid for if the usage behind it goes away. Changing what you run, and whether you run it at all, needs a person who owns it.

The FinOps Foundation warns, in its Automation, Tools & Services capability, that the effort to onboard and understand a tool's data "should not be underestimated", and that a tool's value depends on "the capability and confidence" of the people using it.

Tools can capture discounts and automate actions you approve, and many suggest changes as well, but somebody still has to decide which suggestions are worth acting on, and that happens in a monthly review with the people who own the costs. Buying a tool adds subscription and set-up costs and leaves that management work where it was.

Decisions a Tool Cannot Make

A dashboard can show where the money went. It cannot tell you whose problem it is. I was on a call where a CFO asked his team what it cost to serve their payment gateway to customers. The team had dashboards and more than 90 per cent of their resources labelled, and they could not answer. The dashboards were working; nobody had agreed on the business question the data needed to answer.

In practice, I keep finding the same four things missing:

Ownership means one named person answers for each large cost, and the name is written down.
Decision rights mean somebody has the authority to change the spend as well as see it. Seeing a cost and being allowed to change it often sit with different people.
A monthly review where those owners discuss their costs with the people who can change them.
Behaviour is the slowest: engineers checking what a design will cost before they build it, because someone they respect asks.

The recommendation is usually the easy part. Getting it acted on is a management problem.

An owner can judge the spend the way the business judges any other investment. Spend behind a product that is winning customers may deserve to grow, and spend behind one nobody uses should stop, whatever its unit price. That turns the cloud bill into a portfolio of investments with owners and outcomes you can name.

The clearest case I have is an AI start-up spending roughly $2 million a year on the computing power to train and fine-tune its models. A straightforward cost review would have started with cuts. The founders agreed instead to rebuild how the spend was governed and how its economics were modelled. Spend fell 32 per cent with the pace of experiments intact, and at their burn rate that was about eight months of extra runway they did not have to raise.

When a Tool Comes First

For some companies, a tool has to come first:

You run on several cloud providers, or buy AI capacity from more than one AI model provider. No single provider's reports give you one view, and you cannot hold anyone to a cost you cannot see.
Several product teams share one pool of servers, and your provider's reports cannot split that bill between them.

In those cases, sort out visibility first, then put owners on the costs straight away.

For everyone else, I would sequence it this way:

Your cloud provider's own reports.
Named owners and a regular review.
A paid platform, once that review keeps asking questions the reports cannot answer.

Your finance and engineering leads can do the ownership work themselves. The Foundation makes a related point, warning that over-reliance on outside professional services can create dependency and hold back an organisation's own practice.

A Test Before You Sign

Take the largest increase on last month's bill and ask three things of your current setup:

Can you see where it came from?
Can you name the person who owns that line?
Did that person know before the invoice arrived?

If you cannot see where it came from, better labelling or a tool is the first fix. If you can already see the cost but nobody owns it or saw it coming, more visibility will not solve the underlying problem. If all three answers are yes, buy another platform only when you can name the capability, or the manual work, your current setup cannot handle.

This week, before any renewal or procurement meeting:

Write the owner's name beside each of your five largest cost lines.
Book a thirty-minute monthly review with those owners and whoever signs off the cloud budget, where each owner says whether their line is earning its keep.

The lines that stay blank are the ones to fix first, with or without new software.