When a cloud bill climbs faster than the business, the cause is rarely one big mistake. It is a hundred small ones, none of them anyone's job to fix. On one recent engagement we found that 38% of spend was idle or oversized capacity, hiding in plain sight. Here is where it usually lives.
Instances sized for a peak that never comes
Teams provision for the worst case they imagined during design, then never revisit it. Months later, utilization sits in the single digits. Rightsizing to actual usage, and letting autoscaling handle real spikes, is often the single biggest quick win.
On-demand pricing for steady-state work
If a workload runs all day at a predictable baseline, on-demand is the most expensive way to run it. Committed-use discounts and reserved capacity cut that baseline cost, but only with a commitment plan that matches your real usage curve, not a guess.
The things nobody turned off
Idle load balancers, unattached storage, old snapshots, and dev environments running over the weekend. Individually small, collectively significant. Automated detection and clean-up policies stop the slow leak for good.
No tags, so no accountability
Without consistent tagging, no team can see the spend it creates, so no team owns it. Showback turns an abstract company bill into a number each team can act on. Accountability is the cheapest cost-control tool there is.
Make the saving a habit
The reason savings drift back is that the work was a one-off. A monthly FinOps cadence, with finance in the room, keeps the gains. On the engagement above, a 34% cut held a year later because the habit, not just the fix, was in place. That is the difference between cutting a bill once and keeping it down.