Cloud Costs: The Problem Isn't Spending, It's Visibility

Reflections from a talk at the Futurescot Transforming Public Services conference — but the lessons apply everywhere budgets meet cloud bills.

On 8 May 2025, I had the pleasure of speaking at the Futurescot Transforming Public Services — Cloud, Data & AI Conference in Glasgow. The topic was cloud cost optimisation, and while the audience was largely public sector, the conversation is relevant to everyone. Finance teams in retail. IT directors in manufacturing. CTOs at scale-ups wondering why their AWS bill doubled overnight.

So here's the slightly expanded version of that talk — minus the conference coffee and plus some additional thoughts for anyone trying to get value from cloud investments rather than just paying for them.

The Scale of the Problem

Global public cloud spend hit roughly £600 billion in 2024. It's forecast to double by 2028, largely driven by AI workloads. For context, that's not a gentle growth curve — it's a hockey stick with a jet engine strapped to it.

And here's the uncomfortable truth:

• Cloud budgets run approximately 17% over limit on average
• Around 30% of cloud spend is wasted
• 53% of enterprises say they haven't seen substantial value from their cloud investments

That last one stings. More than half of organisations moved to the cloud and are still waiting for the promised benefits. Often because they did a lift-and-shift and called it a day, without optimising or modernising anything.

Here's the thing — cloud doesn't just happen to be cost-effective. You have to make it cost-effective. And that requires knowing what you're spending money on.

Only 30% of companies know exactly where their cloud budget is going. Which means 70% are essentially hoping for the best. That's not a strategy. That's a prayer.

Cost Optimisation Is Not Cost Cutting

Let's get the terminology straight, because this is where most finance conversations go sideways.

Cost optimisation is not about reducing expenses. It's about strategically reallocating spending toward things that actually matter. The question isn't "where can we cut costs?" but "how can we maximise the value of what we're already spending?"

The Home Office — and I reference them because they published this publicly — adopted cloud optimisation practices like rightsizing and shutting off idle resources. They reduced cloud costs by 40% without sacrificing service delivery. The savings were then reinvested in new digital services.

That's not cost cutting. That's value realignment.

Most of the waste came from developer environments running expensive instances with nobody asking whether the financial impact was justified. Classic case of technical freedom without financial awareness. Guardrails solved it.

The Visibility Problem

Here's where most organisations fall over.

You can't optimise what you can't see. And most organisations have terrible visibility into cloud spend. Who's spending what? Which team owns that mysteriously expensive S3 bucket? Why did compute costs spike 300% last Tuesday?

Computer says: ¯_(ツ)_/¯

Effective FinOps — Financial Operations, the practice of bringing financial accountability to cloud spending — starts with observability. Real-time dashboards. Detailed tagging. Alerts when budgets start creeping. The ability to drill down from "we spent £50k on compute" to "this specific Lambda function is costing us £200/day and nobody knows why."

Some organisations started running daily cost reporting during cloud migrations, so leadership could course-correct in real time rather than discovering overruns at year-end. That's the goal: cloud spend becomes a trackable, controllable metric like any other operating expense.

FinOps in Practice

The FinOps Foundation breaks this down into four domains that operate as a continuous cycle. Not a one-time project. Not a quarterly review. A continuous practice.

1. Understand Usage & Cost

This is the foundation. You need:

• Real-time cost and usage visibility
• Comprehensive tagging — and I mean everything tagged
• Budget threshold alerts and anomaly detection
• Dashboards that executives can actually understand

Key questions: Where are we spending the most and why? What percentage of resources are still untagged? What spiked this week?

2. Quantify Business Value

This is where IT and Finance need to be speaking the same language.

• Map spend to budgets, KPIs, and business outcomes
• Forecast and benchmark so finance teams can plan
• Use the data to make informed decisions about where to invest

Key questions: How does cloud spend track against budget? What's the forecast vs. actual for Application X this quarter?

3. Optimise Usage & Cost

Now you can actually do something about it:

• Rightsizing — especially for workloads from lift-and-shift migrations running on instances sized for "just in case"
• Reserved instances and savings plans instead of pay-as-you-go
• Shutting down non-production environments when they're not in use (you'd be amazed how many dev environments run 24/7 for a team that works 9-5)
• Sustainability benefits — using only what you need isn't just cost-effective, it's GreenOps

Key questions: Which resources are oversized or idle? What savings can we unlock with commitments? Are dev environments actually switching off?

4. Manage the Practice

FinOps isn't a tool. It's a practice. It needs care and feeding:

• Define and refine policies and governance
• Evaluate whether your tooling still meets requirements
• Track maturity and review what's working
• Training for teams that don't have the knowledge to optimise

Key questions: Are our policies and workflows actually working? What training or tooling gaps are blocking adoption?

The Tooling Landscape

Right, let me get the crayons out. Here's how different technology layers map to FinOps capabilities:

The key is integration. Data flowing into dashboards, anomalies triggering alerts, alerts creating tickets, tickets driving action. A closed loop of continuous improvement.

What This Means for Your Organisation

Whether you're public sector justifying taxpayer spend, or private sector trying to maintain margins, the fundamentals are the same:

1. Assess your current state. Do you actually know your cloud spend? Who owns it? Can you identify low-hanging fruit — services sorted by cost, unused resources that could be switched off?

2. Build a FinOps function. Someone needs to own this. It doesn't have to be a massive team, but it does need to be someone's job. Establish processes for ongoing optimisation, not just annual reviews.

3. Educate and empower. Your engineering teams need to understand cloud cost drivers. Not to become finance experts, but to make informed decisions about architecture. Training, certification, community engagement — it all helps.

4. Set targets and track. Treat cloud cost efficiency as a KPI. Increase the percentage of tagged resources. Reduce waste by a measurable amount each quarter. Make it visible to leadership.

Conclusion

Cloud adoption without a FinOps practice is a recipe for inefficiency. And inefficiency, at scale, becomes genuinely painful — whether that's budget overruns, shadow IT, or the slow creep of technical debt disguised as infrastructure spend.

The good news? Most cloud waste is fixable. The bad news? It requires discipline, visibility, and a willingness to have slightly awkward conversations about who's responsible for that mystery database that costs £500/month and nobody remembers creating.

If you're in the same boat — trying to get value from cloud investments rather than just paying the bills — I hope this helps. If you're already running a mature FinOps practice, I'd love to hear what's working for you.

And if you're still flying blind on cloud costs? Start with tagging. Seriously. Tag everything. It's not glamorous, but it's where visibility begins.

This post is based on my talk at the Futurescot conference on 8 May 2025. If you were there and have questions, or if you're wrestling with similar challenges in your organisation, I'd be happy to chat.