25 to 35 percent. That's the average share of enterprise cloud budget going to resources that are idle, oversized, or simply forgotten. Not misconfigured in a way that breaks anything — just quietly running at a size or schedule nobody's revisited.

If that number sounds high, it's because most cost visibility tools show you what you're spending, not whether you should be.

Where the money actually goes

Across audits of Azure environments in finance, retail, and enterprise IT, the leaks cluster into a small number of categories:

Non-production compute. Dev and test environments running around the clock instead of on a schedule matching actual usage hours. This is consistently the largest single category of recoverable spend.

Database over-provisioning. Tiers chosen for anticipated peak load that never quite materializes, left untouched long after go-live because revisiting sizing feels riskier than it actually is.

Storage debris. Unattached managed disks, old snapshots, and data sitting in a pricier access tier than its actual usage pattern justifies.

Unused licensing. Assigned seats and subscriptions that outlived the employees or projects they were provisioned for.

Why "just turn things off" doesn't stick

Here's the part that's easy to underestimate: cost optimization has a half-life. Do a single cleanup pass and, within a couple of quarters, spend tends to creep back toward where it started — new environments get spun up without the same discipline, and nobody's watching the drift.

That's why the organizations that sustain their savings run this as a cycle rather than a project:

1. Assess — actual usage data against actual billing, revisited regularly, not a one-time snapshot.

2. Optimize — right-sizing plus the correct pricing model (reserved, savings plan, or spot) per workload type.

3. Automate — tagging, budget alerts, and shutdown policies that don't depend on someone remembering.

4. Monitor — monthly review cadence, because pricing and workloads both shift constantly.

The governance gap is the real problem

Most of what drives cloud waste isn't a technical failure — it's an ownership gap. Engineering optimizes for velocity and uptime; finance sees the bill without the infrastructure context; and in most organizations, nobody sits at the intersection with both visibility and authority to act.

This is the actual premise behind FinOps as a discipline: it's not a tool, it's an accountability structure.

Worth a second look

If your last cost review was more than six months ago, there's a very good chance meaningful savings have quietly reaccumulated since then — that's simply how cloud environments behave. Cloud 9 Infosystems offers a free assessment that maps exactly where the leaks are in your specific environment before recommending anything.

If you're running a similar cost governance process internally, I'd be interested to hear what's worked (or hasn't) for keeping optimizations from eroding over time.