Why Your Cloud Bill Keeps Growing (and How to Cut It Without Breaking Anything)

The cloud bill arrives. It's up again - not dramatically, a few percent, the same as last month and the month before. Someone asks why. Someone else says it's growth. Nobody's quite sure.

That's how most cloud bills behave. They don't jump; they creep. And when you finally look closely, a surprising share of the money is going on things nobody is using.

Why cloud bills creep

The cloud makes it very easy to start things and very easy to forget them. A server spun up for a test. A copy of the database for a project that ended. Storage that only ever grows. None of it shows up as a decision anyone made - it just keeps running, and keeps charging.

A few habits make it worse:

• Nobody owns the bill. It lands in finance, who can't tell what's needed, while the engineers who could tell never see it.
• Everything runs all the time. Development and test environments that are only used in office hours run nights and weekends too.
• Steady usage is paid for at the highest price. Pay-as-you-go is the most expensive way to buy something you know you'll use all year.
• Storage and logs only grow. Nothing is ever deleted, because deleting feels risky and keeping costs a little more each month.

Where the money usually goes

When you go through a bill line by line, the same leaks turn up again and again:

Servers doing almost nothing. Left over from projects, tests or people who've moved on. They cost the same whether anyone uses them or not.

Servers bigger than they need to be. Sized for a launch, a peak or a guess, and never revisited. If a server never uses more than a third of its capacity, you're paying for the other two thirds.

Non-production running 24/7. If your development and test environments are only needed twelve hours a day, five days a week, they're idle for 108 of the week's 168 hours. Scheduling them off saves around two thirds of their cost, and nobody notices.

No commitment on steady usage. All the big providers discount usage you commit to for one or three years. If some of your servers and databases run all day, every day, paying the pay-as-you-go rate for them is leaving money on the table.

Old snapshots, unattached disks and forgotten storage. Backups older than anyone's retention policy. Disks left behind when a server was deleted. Data nobody has read in years, on the most expensive storage tier.

Logs kept forever. Logging left at full detail after a debugging session, and kept with no time limit.

Things nobody owns. Resources with no owner and no description. Usually the hardest to cut, because nobody is sure what will break - which is exactly why they need an owner.

How to cut it without breaking anything

The goal isn't to slash the bill in a week. It's to stop paying for what you don't use, without taking anything important down with it.

1. Give every line an owner. Every service, account and project on the bill should have a name next to it: someone who can say whether it's needed. Most providers let you tag resources so their cost can be traced to a team or project. Untagged spend is where waste hides.
2. Start with the quick wins. Schedule non-production off outside working hours. Delete unattached disks and snapshots past their retention. Set retention on logs. These are low risk and often take a few hours.
3. Check before you delete. Ask the owner first. If there's any doubt, take a snapshot, switch it off rather than delete it, and wait a couple of weeks to see if anyone shouts.
4. Right-size, then commit. Shrink oversized servers first, so you don't lock in a commitment on capacity you don't need. Then commit only to the steady baseline you're confident about - a one-year commitment on part of your usage is a sensible start.
5. Look every month. A short monthly review of the bill, with the owners, keeps it from creeping back. A one-off clean-up without that habit usually undoes itself within a year.

What not to do

• Don't cut blindly. A big cost-cutting push that takes down production costs far more than it saves, and makes everyone nervous about cutting anything again.
• Don't buy three-year commitments on things you're about to change. If you're planning to move, rebuild or retire a system, wait.
• Don't treat it as an IT problem alone. The decisions - which environments are needed, how long data must be kept, how much resilience is worth paying for - are business decisions. Finance and the technology team need to make them together.

The question to ask your team

If you're not the one who reads the cloud bill, ask whoever does:

"What are the five biggest lines on our cloud bill, who owns each one, and what would it take to cut it by a fifth?"

If nobody can answer the first two parts, that's your starting point. The third part usually turns out to be easier than anyone expected.

The bottom line

Cloud waste isn't a sign of a careless team. It's what happens when spending is easy to start, nobody owns it, and nobody looks. The fix is unglamorous: owners for every line, the quick wins first, and a monthly look at the bill.

Most companies that do it find they've been paying for a lot of things they weren't using.

Want to find yours? The Cloud Cost Checkup takes you through your bill line by line: 16 places cloud bills leak, the saving from each fix, what commitments would save, and the quick wins to do first.

Free guide: The 45-Minute Tech Spend Checkup - a one-page meeting agenda for finding out whether your technology spending is working.

Related reading

• The 45-Minute Tech Spend Checkup Every CEO Should Run
• The "Later" Tax: What Deferred Technology Decisions Really Cost

Not sure what your cloud bill is paying for? Let's talk. A fresh pair of eyes on the bill usually pays for itself in the first month.