How-to guide

Set budgets and cost anomaly alerts on AWS

Set AWS Budgets and Cost Anomaly Detection so you catch overspend early, with concrete steps, checks and pitfalls for small teams.

By the Cloudverg engineering team · · 6 min read

Why this matters

A bill surprises you at month end. By then the spend already happened, and your only move is to explain it. Budgets and anomaly alerts move that discovery forward, often to the day the spend starts. You get a chance to switch off the runaway job or move the workload before the next invoice.

AWS Budgets tracks your actual or forecasted spend against a threshold you set. Cost Anomaly Detection watches for spend that does not match your normal pattern, even when it stays under a fixed budget. The two do different jobs. Budgets answer "am I over my plan?" Anomalies answer "is something behaving strangely?" You want both.

Alerts only help if someone owns them and acts. Decide now who reads the email or Slack message, and what they are expected to check first. An alert nobody triages is just noise in an inbox.

Keep the setup small at first. One budget per account per environment, one anomaly monitor for the whole account, and one place where alerts land. You can tighten it later once you see what real spend looks like.

Before you start

You need the right permissions. In AWS Identity and Access Management the managed policy Billing gives read access to billing data, and adding Cost Anomaly Detection actions lets someone create and edit monitors. Only accounts in the management account of an AWS Organization or a standalone account can see consolidated billing data. Member accounts cannot create budgets that cover the whole organization.

Turn on Cost Explorer first, even if you only use it briefly. Cost and usage data feeds both Budgets and Anomaly Detection, and it takes up to a day to populate after you enable it. If you are new to the account, the first anomaly results need some history before the model has a baseline.

Decide your structure. Budgets can filter by linked account, tag, service or cost category. If you do not have reliable cost allocation tags yet, start with budgets per linked account and per service. That needs no tagging discipline and still tells you which account drifted.

Collect the addresses that will receive alerts. Set up an Amazon SNS topic in the same Region as your budget, then subscribe the email addresses or the chat integration your team reads. Confirm the SNS subscription, because an unconfirmed subscription silently drops messages.

How to do it

Start with budgets that match how you think about money. A full-account monthly budget catches total drift. Per-environment or per-service budgets catch the one team or feature that went sideways. Set the amount from your recent actual spend, not a wish. A budget set far above reality never fires.

Add a forecasted threshold as well as an actual one. Actual spend tells you what already happened. Forecast tells you where the month is heading while there is still time to react. A forecast alert at 100 percent of budget gives you a warning days early.

Then set up anomaly detection. AWS builds an independent monitor per service and per linked account by default. You can narrow it with filters, for example to watch only EC2 and RDS, or only accounts that run production. Set the alert threshold so it does not fire on every small blip, then route it to the same SNS topic so alerts land in one place.

Keep the alert messages useful. Include the budget name or monitor name, the account, and the top contributing service. Whoever gets the alert should know where to look without opening five screens.

  1. Enable Cost Explorer. Open Billing and Cost Management and turn on Cost Explorer. Wait up to a day for data to appear before you rely on budgets.
  2. Create an SNS topic for alerts. Create a topic in the Region you use for billing and subscribe your team's email or chat endpoint. Confirm every subscription.
  3. Create a monthly budget. In AWS Budgets, choose a cost budget, set the period to monthly and enter an amount based on recent actual spend. Name it clearly, such as prod-account-monthly.
  4. Add actual and forecasted thresholds. Set an actual spend alert at 100 percent of budget and a forecasted alert at 100 percent. Attach the SNS topic to both.
  5. Add filter-based budgets. Create a second budget filtered by linked account, tag or service for the workload you care most about. Repeat for each environment you track.
  6. Set up Cost Anomaly Detection. Open Cost Anomaly Detection and create a monitor for the account or service you want watched. Use filters to cut noise from low-impact services.
  7. Configure the anomaly alert. Set the alert threshold so small blips do not fire, and route it to the same SNS topic. Add the monitor to your on-call runbook.
  8. Test with a small budget. Create a temporary budget with a tiny amount and confirm the alert arrives. Delete the test budget once you see the message.

How to check it worked

Do not wait for a real overspend to test. Create one temporary budget with a tiny amount you will definitely exceed, using the actual spend threshold. Confirm the SNS email arrives and that the subject and body name the right account. Delete the test budget afterwards.

Check the AWS Budgets console for budget status. Each budget shows its current and forecasted spend against the limit, and the history graph shows when it crossed a threshold. If the graph is empty, your filter probably matches nothing, or cost data has not populated yet.

For anomalies, look at the Cost Anomaly Detection page and open the monitor. You should see a list of detected anomalies with impact and root cause. If you have not had one yet, that is normal for a quiet account. The monitor still shows as active, which is what matters.

Verify the SNS subscription is confirmed for every address. An unconfirmed subscription is the most common reason someone "never got the alert". Check the topic's subscriptions tab and resend confirmation where needed.

Mistakes to avoid

Setting budgets without filters. A single account-wide budget is useful, but it will not tell you which team or service caused the drift. Add filter-based budgets for the dimensions you actually manage, and keep the account-wide one as a backstop.

Ignoring forecasted thresholds. If you only alert on actual spend, you learn about overspend after it is spent. Forecast alerts buy you days of warning and cost nothing extra.

Letting anomaly sensitivity stay default without checking the noise. If every deploy triggers an alert, people stop reading them. Adjust the threshold or narrow the monitor to services where a surprise really hurts.

Forgetting that budgets are not enforcement. A budget alert does not stop spend. If you need a hard stop, look at AWS Budgets actions, which can apply an IAM policy or stop instances, and test them carefully. Also remember that credits, refunds and Savings Plans can change how actual spend compares to your plan. If your alerts feel unreliable, a Cloud cost audit at $4,500/month is one option among others to review the whole picture.

Frequently asked questions

What is the difference between AWS Budgets and Cost Anomaly Detection?

Budgets compare spend against a limit you set, so they catch planned overspend. Anomaly Detection watches for unusual patterns even when spend stays under budget. Use both, because each catches what the other misses.

Why am I not receiving my AWS budget alerts?

The usual cause is an unconfirmed SNS subscription, or alerts routed to an email nobody checks. Confirm the subscription in the SNS console and verify the budget action points to the correct topic. Also check that your budget filter matches real spend.

Can I set a hard cap on AWS spending?

Budgets and anomaly alerts report, they do not stop spend by themselves. AWS Budgets actions can apply an IAM policy or stop instances when a threshold is crossed, but they are not a true spending ceiling. Test any action carefully before relying on it.

Monthly plan

Want help with this? Cloud cost audit

$4,500/month

2 weeks · Starts within 10 business days

A two-week, read-only review of your AWS, Azure, Google Cloud or Oracle bill, with every saving ranked by dollars per month.

Renews monthly. Cancel any time.
What is included

Related guides

More cloud costs guides   Cloud cost audit

Smallest way to start

Get a senior architect on your problem this week

Ninety minutes a month on the cloud, AI or security question that is blocking you, a pre-read of your material and a written summary within two business days.

$450/month

Renews monthly. Cancel any time.