The 30% problem: cloud waste is an execution failure
The Lizrd team · · 2 min read
Every cloud bill has waste in it. Industry estimates put cloud waste at 20–30% of spend — over-provisioned instances, idle databases, orphaned volumes, environments nobody turned off. What’s striking is that the number has barely moved in a decade, despite an explosion of cost-visibility tools.
If dashboards fixed waste, this problem would be solved. It isn’t. So it’s worth asking why.
Seeing isn’t fixing
Cost tools are very good at one thing: showing you where the money goes. You can slice spend by service, by team, by tag, by day. That’s genuinely useful — but it stops at a chart.
A chart doesn’t tell an engineer which node group is safe to shrink, by how much, or what happens if the change is wrong. It hands them a number and a research project. And research projects lose, every time, to shipping features and keeping things up.
The execution gap
The gap isn’t awareness. Most platform teams can tell you, roughly, that they’re overspending. The gap is between “we know there’s waste” and “the change is merged.”
Closing it requires three things a dashboard can’t give you:
- A specific change, not a category. Not “compute is 42% of your bill” — the exact instance type to switch to, on the exact resource.
- Confidence it’s safe. The evidence (30-day utilization), the risk, and how to roll it back.
- Proof it worked. Did the bill actually go down, or did the saving evaporate?
What closing it looks like
This is the whole idea behind Lizrd: treat optimization as an execution problem, not a visibility one. Find the highest-impact savings, hand the team the exact fix as a copy-paste diff, and validate the realized savings afterward.
Waste doesn’t persist because it’s invisible. It persists because fixing it has been too much work. Make the fix cheap, and the 30% finally starts to move.