FinOps doesn't start with a platform. It starts when someone can explain, with evidence, why the bill went up and who can change it.

Before you optimize, get a baseline

Take the last three months and break spend down by account, environment, service and owner. If your tags aren't consistent, that isn't a reporting detail: it's the first sign of operational debt.

Rule of thumbIf nobody can explain 20% of your bill, you don't need more dashboards yet. You need ownership.

The three waste signals that pay off most

Idle capacity

Look for EC2 with sustained low CPU, unattached EBS volumes, unused public IPs and environments that outlive business hours.

Discounts without demand

A Savings Plan only optimizes a stable workload. Weigh commitment, utilization and expected growth; a discount isn't a saving if it leaves you overprovisioned.

Architecture nobody discusses

RDS, NAT Gateway, data transfer and logs tend to grow with no product ticket to justify them. Put them on the monthly review.

The minimum monthly ritual

01Measure

Bill, usage and anomalies against the previous month.

02Decide

One owner per action, with a date and an expected saving.

03Learn

Write down what changed and what won't happen again.

When this ritual works, a tool adds speed. When it doesn't, a tool just adds another view nobody opens. Estimate your opportunity range

Methodology note

This guide is educational. Its recommendations don't replace an analysis of your account, contract or architecture.