FinOps for startups: a 30-day plan without a FinOps hire

FinOps is the practice of making cloud spending a shared, informed decision between the people who build and the people who pay. At a large company that takes a team. At a startup it takes an owner and a routine: someone who can say what the cloud cost last month, why it changed, and what is worth doing about it — in under half an hour a week.

The owner does not have to be a finance person. At most startups it is a founder, the CTO or the platform lead — whoever already gets the question when the bill surprises someone.

1. 1Turn the data on. Open Cost Explorer in the Billing console, and set up a Cost and Usage Report export through AWS Data Exports to an S3 bucket, so you keep line-item history from now on.
2. 2Pick one cost basis and keep it. If you have Savings Plans or reservations, use amortized cost, so an upfront payment is spread over its term instead of looking like a spike. Report every number on that basis.
3. 3Set two alarms. An AWS Budget on total monthly cost, alerting on actual and forecast spend, and AWS Cost Anomaly Detection, which is free. Both take minutes and catch the surprises between reviews.

1. 1Agree on three tags. Team (or owner), service and environment. Set them in your infrastructure code, not by hand, so new resources arrive tagged.
2. 2Activate them for billing. User-defined tags do not appear in cost reports until you activate them as cost allocation tags in the Billing console. It is the step most teams miss.
3. 3Separate production from everything else. Separate AWS accounts for production, staging and experiments — AWS Organizations has no charge — so an experiment can never hide inside production spend.
4. 4Keep untagged spend as its own line. It is the to-do list for your tagging, not noise to spread across teams.

One agenda, every week, with the same people:

• What changed: this week against last, on the same cost basis.
• The top three movers, each with a named owner.
• For each one, a decision: fix it, accept it (it is growth you want), or watch it for another week — written down.

The written decision is the part that compounds. After a month you have a record of what the cloud costs, why, and who decided — which is most of what a board, an investor or a future FinOps hire will ask for.

Rank candidates by how much they would save, how sure you are, and the effort and risk they carry. The usual first wins at a startup:

Count a saving as real only when the bill shows it. An estimate is a reason to act, not a result.

• Total cloud cost this month, and the change against last month — explained, not just shown.
• Cost per team or product, with the untagged share stated next to it.
• When you can measure it, cost per customer or per unit of what you sell: the number that says whether growth makes each customer cheaper or dearer to serve.
• Commitment coverage, once you have commitments.

The signs: spend spread across many accounts or more than one cloud, teams that need their own budgets and showback, AI or GPU costs that need a margin per customer, or a review that no longer fits in thirty minutes. That is when a dedicated FinOps hire or a platform starts to earn its cost.

CloudJaeger is built for this stage: engineering-led SaaS teams, mostly on AWS, without a FinOps hire. It connects through a read-only IAM role, explains what changed in the bill and who owns it, keeps untagged spend visible as its own line, and ranks the fixes worth an engineer’s time — with every saving labelled projected until your bill confirms it. AWS runs on real bills today; Azure and Google Cloud sync real cost and inventory, and stay in Preview until checked against a real bill.