Ad · 728×90 desktop · 320×50 mobile

Why your cloud bill doubled, and what to do about it

Published 2026-04-24 · FinOps & Cloud Cost

Cloud bills don't double by accident — they double because of a specific, identifiable event or drift pattern. If your bill just doubled, you're looking at one of five causes. Start here instead of panicking.

1. A new workload shipped without cost allocation

The single most common cause. An engineering team launched a service, nobody tagged it, and the spend lands in "untagged". Check the billing console's untagged slice over the last 30 days. If untagged doubled, you've found your answer.

2. An auto-scaling event that stuck

Traffic spiked, the autoscaler scaled up, but the scale-down path was misconfigured or slower than the scale-up. Check max-concurrent instance counts by service over the past 30 days. Anything that climbed and plateaued — rather than climbed and returned to baseline — is your target.

3. A data transfer spike

Data transfer (egress) is the silent bill killer. Common causes: cross-region chattiness from a new service, a backup job that shipped outside the intended storage tier, or an application misconfigured to use a public endpoint instead of a VPC endpoint. The data transfer line item in your billing breakdown is where to look.

4. A commitment expired

If you had Reserved Instances or Savings Plans that expired, on-demand rates kick back in and your bill can jump 30–50% for the affected workloads overnight. Check your commitment coverage over time; a gap coincides with the bill spike.

5. A logging or observability tool going feral

A badly-configured Datadog / New Relic / Splunk ingestion can add $10–50k/month. Look at per-service log volumes before and after the spike.

What to do in the next hour

Run the Idle Resource Finder and Rightsizing Recommender in parallel. These won't find the cause of the spike, but they'll free up budget while you investigate, giving you breathing room.

What to do in the next week

Do a Cloud Cost Audit and share findings with engineering. The point is not to assign blame — it's to agree on monthly cost review rituals so the next spike is caught in days, not at the end of the month when the bill arrives.

Ad · 300×250 responsive

Joined by engineers at enterprises running the systems you build. No spam. No tracking in the emails.