When organizations move to the cloud, the promise is always the same: flexibility, speed, efficiency. Pay only for what you use. Scale as you grow. Stop worrying about infrastructure and start focusing on your business.
Then the bill arrives. And it keeps arriving. And it keeps growing, often regardless of whether the business is growing with it.
The uncomfortable truth is that cloud overspending is rarely a technical problem. It’s a decision problem. And the decisions that drive it are often years old, buried in architecture choices nobody remembers making.
To dig into this, we brought together Nikola Djordjevic, Solution Architect, and DevOps Architects Iulian Constantinescu, Marko Prokic, and Dragan Skrinjar – four our practitioners who spend their days building, maintaining, and optimizing cloud environments for real clients. The conversation surfaced three patterns that show up again and again in organizations struggling to get their cloud costs under control.
Your cloud bill is a mirror of your architecture
The most common mistake organizations make is treating a high cloud bill as a financial problem to be managed, rather than an architectural signal to be read.
When leadership sees a large invoice, the natural response is to demand cost-cutting or launch an optimization project. But as Iulian Constantinescu put it during the webinar: “High spending actually masks the real design issue, and teams get used to throwing more instances, bigger databases, or extra storage at problems instead of fixing the root cause.”
The result is a self-reinforcing loop: technical debt increases costs, and high costs consume the time and focus needed to reduce the debt. Legacy patterns lifted and shifted to the cloud, quick solutions built under pressure, over-provisioned resources – these aren’t cloud problems. They’re design problems that the cloud makes more expensive.
The recommendation from the Levi9 team: stop treating cloud cost optimization and technical debt reduction as separate initiatives. They are two sides of the same problem, and solving them together is how you break the cycle.
If everyone owns the cloud, no one owns the costs
The second pattern is ownership, or the lack of it. Cloud environments are almost always shared across multiple teams and services. But while the bill is centralized, responsibility rarely is.
Marko Prokic walked through a four-block framework the team uses to fix this: a clear tagging strategy that maps every resource to an owner, team, or service; cost allocation that ties spending to specific teams or budgets; proactive budgets and alerting that bring teams into the conversation before costs spiral; and visibility through dashboards that give teams a real view of what they’re spending.
The difference this makes in practice can be dramatic. Marko shared an example from a client engagement: “After the process of modernization, when we started looking at their costs and how the services were architected and deployed, we noticed there was a bug – with multi-layer encoding implemented. There was a one-click fix which actually saved them $7,000 a month only for that one endpoint.”
That bug existed before the modernization. It only became visible – and fixable – once the team had genuine ownership of and insight into their own costs.
Waste doesn't accumulate - it's provisioned and forgotten
The third pattern is lifecycle management. Or rather, the absence of it.
Most organizations are good at provisioning cloud resources. They are significantly less good at decommissioning them. Resources get created for a project, a test, or a temporary workload, and then they keep running. Month after month.
Iulian described a scan of one client’s environment where logging and data systems had been accumulating data for at least 15 years, well beyond any legal retention requirement. The fix was straightforward: “Just applying lifecycle management on those resources basically reduced the monthly bill around 20% in production and around 17% in the other non-production environments.”
The principle is simple: every cloud resource should have a defined lifecycle — from provisioning with clear ownership, through active usage and optimization, to automated decommissioning when it’s no longer needed. Automation is the real enabler here. Policies for automatic cleanup, scheduled shutdowns for non-production environments, and intelligent scaling can eliminate enormous amounts of waste without ongoing manual effort.
Practical takeaways
- Connect the financial conversation to the technical one.When Dragan Skrinjar works with clients, the question he finds most usefulisn’t “what’s technically wrong?” – it’s “what is costing us money every single month?” Framing technical debt in financial terms creates a shared language between engineering and finance, and makes it much easier to prioritize what to fix first.
- Build cost awareness into engineering,without building bureaucracy.Cost control shouldn’t mean that every infrastructure decision needs financial approval. That slows delivery and kills autonomy. The goal is to make cost-awareness a natural part of how engineers design and deploy — through clear policies, automated alerts, and good tooling, rather than approval gates.
- Start with an assessment, not an optimization project.Across thewebinar, the consistent message was: most organizations know there’s a problem, they just don’t know where to start. A structured assessment of your cloud workloads, identifying where waste is concentrated, where ownership is unclear, and where the architecture is working against you – gives you an actionable roadmap rather than a list of generic recommendations.
The path forward
One client the team worked with started with an on-premise data center and went through a two-phase journey: a lift-and-shift migration followed by a full modernization to microservice and serverless architecture. The process took roughly a year and a half. At the end of it, they were paying 30% less than they had for the original data center.
Cloud cost control isn’t a one-time project. It’s an ongoing discipline – one that treats every resource as a business asset with a defined lifecycle, and every architecture decision as something that will eventually show up on a bill.
If cloud costs are on your agenda, The Cost of Choice webinar recording is available to watch in full. And if you’d like to explore what this means for your specific environment – our team is ready to talk.
