Free calculator
Is my AWS bill too high?
Most founders ask this and very few have anything to compare against. Put your spend next to typical ranges for your team size and see where you land.
Last month's invoice total is fine.
People who ship to production — a rough proxy for workload scale.
Spend per engineer
$666.67/month
High side
Typical recoverable waste
$1,600.00/month
$19,200.00/year · ≈ £1,248/month · 20% — our audit guarantee threshold
Above the usual range for your team size. Sometimes justified (data-heavy products, ML workloads, high traffic) — but at this level, idle instances, oversized databases and zero Savings Plan coverage usually account for a meaningful slice.
A rough heuristic rather than a law. The bands come from published SME cloud spending benchmarks, not from a portfolio of our own clients, and data heavy or ML products legitimately sit above them. The per engineer lens works because engineers are the people who create infrastructure, and it stops working for companies whose product is infrastructure.
A ratio can only tell you whether your bill looks unusual. Four questions on the estimator will break it into named opportunities with the arithmetic shown, and the free scan after that reads the actual account.
Estimate the whole billWhy bills drift high (even in careful teams)
- →Nothing on AWS turns itself off. Every resource keeps billing until a person stops it. Careful teams create resources carefully and then forget about them at the same rate as everyone else.
- →Growth hides waste. When revenue is growing, a creeping AWS bill reads as "scaling". The waste compounds quietly inside the growth curve and only becomes visible when someone finally looks.
- →Nobody owns the bill. Between ten and eighty people there is rarely a FinOps function. The CTO sees the total, each engineer sees their own services, and the idle m5.2xlarge in eu-west-2 belongs to nobody.
- →The defaults favour AWS. gp2 rather than gp3, logs that never expire, on demand rather than Savings Plans, flow logs capturing everything. Each default is safe and convenient, and each one is the more expensive option.
Frequently asked questions
What is a normal AWS bill for a small company?
As a rough band, UK product companies commonly land between $150 and $400 per engineer per month. A ten person engineering team spending $2,000 to $4,000 a month is unremarkable. The same team at $10,000 a month either runs something data heavy or is carrying real waste. Team size is only a proxy, and the calculator says where it stops working.
Why benchmark by engineer count?
Engineers are the people who create infrastructure: services, environments, experiments. More engineers means more resources, including the ones everyone forgets. Revenue and traffic matter too, but engineer count is the number a founder can give you without looking anything up.
My bill grew but traffic did not. Why?
Usually storage accumulation first, because snapshots, logs and S3 keep growing whether traffic does or not. Then instances and environments left behind by finished projects, price changes such as the 2024 public IPv4 charge, and expired Savings Plans quietly putting everything back on on-demand rates.
How much of a typical AWS bill is waste?
Published industry estimates put 20% to 35% on accounts that have never had a structured cost review, in idle compute, oversized instances, unattached storage, logs that never expire and no commitment coverage. We have not audited enough accounts to add our own figure to that yet. What we can tell you is that our own account came in at 3.4%, and the full result is published on the demo page.
Is this benchmark accurate for my specific company?
No, and it is not meant to be. It is a thirty second sanity check. Data pipelines, ML training and video workloads all sit legitimately above these bands. The only honest answer for a specific account comes from reading that account, which is what the free scan does, read only, in a couple of minutes.