The Automation Illusion: Why ‘Fully Automated’ Doesn’t Mean Optimized 

Part 2 of our Inside the Execution Gap series 

Ask any FinOps leader where they stand on automation, and you’ll probably hear a confident answer.  

In the 2025 CloudBolt Industry Insights (CII) Report, 93% of senior IT leaders said they’re using automation for cost optimization. At first glance, it sounds like a win.  

But the data tells a different story. 

• 65% said at least 20% of optimization recommendations go unimplemented. 
• Most teams still take weeks or longer to remediate waste—even with automation in place. 
• And outside of routine shutdowns, few are optimizing more complex workloads at scale. 

So why aren’t those automation investments delivering results? 

In Part 1 of this series, we explored the risks of expanding FinOps scope too soon. Here in Part 2, we unpack the difference between perceived automation and actual optimization—and why so many teams are still stuck in what we’re calling the automation illusion. 

Everyone Says They’re Automated. Few Actually Are. 

When we asked IT leaders whether their organizations use automation for cloud cost optimization, the answer was a near-unanimous yes. 

But when we dug into what “automation” actually meant, the scope narrowed fast. 

• Shutdown schedules and idle resource cleanup were by far the most common. 
• Kubernetes rightsizing, commitment-based discounting, and real-time remediation were far less prevalent. 
• And only a third of teams said they consistently automate purchasing decisions tied to optimization goals. 

This disconnect reveals the problem: “automation” is being defined far too loosely—often as nothing more than scripted schedules or alert-based workflows.  

As Tracy Woo, Principal Analyst at Forrester, pointed out in our recent webinar, Beyond the Dashboard: Why FinOps Must Evolve for AI, Automation, and Reality: “I have a lot of suspicion around what’s being labeled as automation. If you’re talking about scheduled shutdowns or idle resource scripts that still require a human to say yes, that’s not automation. That’s just orchestration. True automation means no manual intervention. The system should understand when resources need to be on or off—and act automatically, without someone approving it first.” 

The Real Problem: Insight Without Action 

The issue isn’t that FinOps teams lack data or tooling. 

The insights are there. The dashboards are full. 

But visibility doesn’t reduce spend. Action does—and that’s where the breakdown happens. 

Most teams are stuck in a fractured two-step process: 

1. FinOps surfaces an opportunity (idle resource, misconfigured service, underutilized commitment). 

2. Engineering or operations re-validate the opportunity, re-prioritize it, chase down approvals…and eventually deploy a fix—often weeks later.  

By the time a single optimization is implemented, the opportunity may have vanished—or the delay has cost the business thousands. 

As CloudBolt CTPO Kyle Campost explained during the webinar: “The full insight-to-action loop—connecting waste signals all the way through evaluation, prioritization, and execution into production—is almost never automated. And that’s the disconnect. High-performing engineering teams measure deployment lead time in hours. But FinOps teams often wait weeks for changes to go live. That’s a gap wide enough to kill momentum—and real savings.” 

What True Optimization Maturity Looks Like 

To escape the automation illusion, teams need more than alerts and scripts. They need a system that closes the loop—and keeps it closed.  

Here’s what that looks like in practice:  

• Close the loop: Optimization doesn’t end at detection. It ends when the change is deployed in production. 
• Measure lead time: Not alert-to-ticket. Alert-to-action. How long does it really take to fix an issue? 
• Minimize approvals: Design trust boundaries that allow safe, repeatable optimizations to proceed automatically.  
• Empower engineers: Give the teams closest to the code the visibility and autonomy to act—not just react to FinOps escalations. 

Not sure where you stand? Start here: 

• Are we automating action—or just creating better alerts? 
• How long does it take to remediate waste? 
• Can engineers fix cost issues without waiting on approvals? 

If the answers are uncomfortable, that’s not failure. It’s clarity. And that’s the first step toward maturity.  

Coming Up Next: Proving ROI 

In the final part of this series, we’ll tackle the question that looms over every FinOps initiative: Are we actually proving ROI? 

Because without clear, trusted links between cost optimization and business value, even the most efficient teams are flying blind.  

Until then, download the full 2025 CloudBolt Industry Insights Report to explore all the findings behind the execution gap.