CloudCostCompareHidden cloud costs and where they show up on the bill
Short answer: most surprise cloud bills do not come from the big line items. They come from small meters nobody put in the plan: data moving between services, gateways that bill by the hour, logs, forgotten disks and backups. They hide because a server has a price on its page and a gigabyte moving between two services has a price on a different page.
Key points
- The largest hidden lines are usually data transfer, NAT, logs and idle resources.
- Estimates built from a server list leave these out.
- Stopped machines still bill for disks and addresses.
- Find them by sorting the bill by service and by tag.
- Alerts on each hidden category catch growth early.
What are the common ones?
| Cost | Why it hides | Where to look |
|---|---|---|
| Data transfer | Billed per gigabyte leaving the cloud or crossing zones, not part of the server price | Network or bandwidth lines. See egress |
| NAT and private networking | A gateway charge by the hour plus per gigabyte processed | NAT gateway costs |
| Idle resources | Stopped servers still bill for their disks and addresses | Compute, disk and IP lines for resources with no traffic |
| Logs and metrics | Verbose logging is billed by volume ingested and retained | Logging and monitoring costs |
| Backups and snapshots | Kept long after the system they protect | Backup and retention costs |
| Load balancers | An hourly charge even with no traffic | Load balancer costs |
| Support plans | A share of spend or a monthly fee | Support plan costs |
| Request charges | Millions of tiny reads and writes on storage and queues | Object storage pricing |
| Security add-ons | Firewalls and DDoS plans billed per rule or per month | WAF and DDoS costs |
What does a worked example look like?
Made-up round numbers to show the method, not real prices. A team estimates a small web app from the server list.
| Line | In the estimate | On the bill (made up) |
|---|---|---|
| 2 app servers | 2 x 40 = 80 | 80 |
| Database | 120 | 120 |
| Load balancer | 20 | 20 |
| NAT gateway: 730 h x 0.05, plus 500 GB x 0.05 | not included | 36.50 + 25.00 = 61.50 |
| Logs: 300 GB ingested x 0.50 | not included | 150.00 |
| Idle test disks and a forgotten snapshot | not included | 18.00 |
| Support: 10% of spend | not included | about 45 |
| Total | 220 | about 494 |
The estimate was off by more than half, and each missing line is small alone. The biggest, logs, is a setting (log level), not a service. That is why hidden costs are found by reading the bill by meter, not by looking at the servers.
Why are they missed?
A server has a price on its page. A gigabyte moving between two services has a price on a different page, under a different heading. Estimates built from the server list leave the second kind out.
How do I find mine?
- Sort last month’s bill by service and read the top ten lines.
- Group by tag to see which project owns each. See tags and labels.
- Look at lines that grew faster than traffic did.
- Delete or schedule what nothing uses. See dev and test environment costs.
- Set an alert on each hidden category. See budgets and cost alerts.
Which mistakes make it worse?
- Debug logging left on in production.
- One NAT gateway per environment that nobody uses.
- Snapshots with no expiry date.
- A support plan on every account, including tests.
- Copying data between regions on a schedule that nobody needs.
Decision checklist
- Does my estimate include data out, NAT, logs, backups and support?
- Is every resource tagged with an owner?
- Is there an alert on each of the top five meters?
- Do old snapshots and test environments have an expiry?
Next steps
Our estimates include the extras we can size from public prices, and show a gap, not zero, for those we cannot. See how to read a cloud estimate, how to read a cloud bill and the monthly optimization checklist.
Frequently asked questions
What are the most common hidden costs in the cloud?
Data transfer, NAT gateways, idle resources such as disks and IP addresses, log ingestion, old backups and snapshots, load balancers that run with no traffic, support plans and request charges on storage and queues.
Why is my cloud bill higher than the estimate?
Estimates are usually built from the server list, and the extras live on different pages with different meters. Compare the bill by service with the estimate line by line to see which meter was missed.
How do I find unused cloud resources?
Sort the bill by resource, look for lines with cost but no traffic, and list disks, snapshots, IP addresses and load balancers that are not attached to anything running.
Do stopped virtual machines still cost money?
The compute charge usually stops, but disks, snapshots and reserved IP addresses keep billing. Check the vendor's wording for stopped versus deallocated machines.
How can I stop surprise cloud bills?
Set budgets with alerts per account or project, tag resources by owner, review the top ten bill lines monthly, and delete or schedule what nothing uses.