Cloud cost optimisation for maximising value

Why cloud optimisation matters

Cloud optimisation is essential for achieving a balance between innovation and financial efficiency, especially as cloud spending continues to rise. Studies show that 30% of cloud spending is often wasted, while organisations implementing effective cloud cost optimisation strategies can save an average of 20-25% on cloud expenses.

As cloud technologies evolve, so do optimisation practices and frameworks. The need for scalable solutions and the growing use of advanced technologies, such as generative AI, mean that cloud investments now go beyond core infrastructure to include data analytics, machine learning, and large-scale innovation. This broader scope makes disciplined cost management vital—not only to stay within budget but also to free up resources for strategic initiatives and new technology adoption.

This section examines cloud optimisation as a strategic blend of cost-management techniques, proactive financial planning, and a culture of cost-awareness. Our experienced consultants share best practices for maximising cloud resources, balancing scalability with budget discipline, and fostering a sustainable cloud strategy. With these insights, organisations can set the stage for innovation and long-term growth, ensuring cloud investments drive meaningful value.

Core cost-optimisation strategies

Effective cloud cost optimisation is about maximising the value of cloud resources, not merely minimising costs. Below are key strategies informed by our experience.

Predictive cost management and proactive monitoring

Predictive tools, such as cost calculators and budget tracking, allow organisations to estimate expenses before deployment. Using cloud vendors’ budget calculators and setting budget alerts can help teams avoid surprise cost spikes. While predictions delivered using these methods often have limitations, combining them with ongoing tracking allows teams to adjust as demand changes, keeping spending aligned with budgets.

Empowering teams for cost accountability

Cloud cost management starts with empowering teams to monitor their resource usage and make cost-efficient decisions. When teams have visibility into their spending, they can make proactive adjustments, preventing the need for reactive cost cuts.

We’ve seen that giving teams autonomy to monitor their own costs prevents last-minute budget cuts. One unique approach involved using a limited credit card for cloud expenses: when the balance ran out, the team was notified and prompted to investigate spending. While simple, this method brought immediate attention to expenses and fostered a habit of frequent reviews.

A cost-conscious culture is essential for sustaining cloud optimisation across development, finance, and operations. Fostering this culture involves collaborative cost goals and encouraging accountability through training and transparency.

Expanding focus beyond major services

While EC2 and RDS often dominate cloud bills, focusing only on these services can overlook other significant expenses that could be optimised. Managed services like Databricks, Managed Kafka, and DynamoDB often carry high costs, and scrutinising these less obvious services can reveal additional savings.

Examining costs beyond the obvious big-ticket items has helped many organisations uncover overlooked expenses. In our experience, taking a detailed view across all services often identifies optimisation opportunities in unexpected areas, yielding greater cost control without impacting critical functions.

Value alignment in cost management

Effective cost management isn’t just about cutting expenses—it’s about ensuring that every dollar spent on cloud services adds value to the business. When evaluating high-cost items, the focus should be on their contribution to critical business functions. Services driving substantial revenue may justify their costs, while high expenses on non-critical services require closer examination.

For example, a high-cost service like DynamoDB might be essential if it supports customer-facing applications or revenue-driving functions. In contrast, non-critical services with high costs might be better optimised or scaled back.

Avoiding overprovisioning through right-sizing and auto-scaling

In our experience, overprovisioning to handle demand spikes—common in fintech—frequently leads to inefficient spending. Instead of overprovisioning, which often leads to reactive cost-cutting cycles, organisations should use right-sizing and auto-scaling to align resources with actual demand. Right-sizing ensures resources match workload needs, while auto-scaling dynamically adjusts resources to accommodate demand spikes without wasting capacity.

Granular cost attribution for organisational accountability

Larger organisations often face challenges with attributing cloud costs accurately across divisions. This lack of visibility can lead business unit leaders to overlook cloud expenses when building business cases, which complicates efforts to optimise overall costs. Without clear cost attribution, departments may use cloud resources freely, leading to overspending and difficulties in identifying areas for cost savings.

Accurate cost attribution enables organisations to assign cloud expenses directly to business units, fostering accountability and informed decision-making. This is particularly important in larger enterprises where shared services are common, and resources need to be optimised collaboratively.

Granular cost attribution remains rare in large enterprises due to the complexity of implementing it. However, in one client example, adopting a detailed cost attribution approach allowed cloud costs to be assigned directly to business units. This traceability provided insights into exact costs per product or service, helping leaders identify areas for cost reduction and make more strategic financial decisions. This practice allowed the organisation to control costs effectively, in contrast to other companies where detailed attribution is lacking, often resulting in unchecked demand from departments without accountability for their cloud spending.

Foundational frameworks like FinOps have become critical in cloud optimisation, helping organisations manage cloud spending through a collaborative approach across finance, operations, and engineering. By promoting accountability and structured tracking, FinOps aligns cloud costs directly with business objectives. This allows organisations to achieve a sustainable balance between scalability and budget, enabling ongoing innovation without financial strain.

It is however, important not to let expense tracking itself become a drain on resources. For larger organisations in particular cost allocation can get complex, especially in hybrid cloud setups. Organisations should focus FinOps efforts on high-cost or high-impact areas, using cost allocation tools where they provide clear value without over-complicating tracking.

Data lifecycle management and intelligent storage tiers

Many organisations, particularly those with extensive data sets, find that “limitless” storage quickly leads to ballooning costs. A data lifecycle policy helps control costs by moving infrequently accessed data to lower-cost storage and deleting data that no longer provides value. It can be hard to convince an organisation to implement data deletion policies, but it’s well worth it. Intelligent tiering and periodic data reviews ensure storage expenses align with the data’s business value.

Aligning cost optimisation with cloud-native architectural principles

Cloud-native architecture is a frequently overlooked yet vital strategy for effective cloud cost optimisation. Legacy applications often get moved to the cloud without re-factoring, less-than-optimal cost and performance outcomes. 

Cloud-native solutions prevent organisations from duplicating on-premises dependencies in the cloud. Containerised applications and serverless functions scale based on demand, reducing the need for over-provisioned resources.

Cloud architectures should evolve as applications scale. Regular reviews ensure that applications remain optimised and leverage the most efficient configurations..

Understanding the cost drivers in new technology applications

New areas like Generative AI present exciting opportunities for innovation, but also bring significant operational and cloud infrastructure costs. For many organisations, the path to deploying Large Language Models (LLMs) is filled with assumptions about model size and performance, often leading to unnecessary expenses. It’s important to understand the cost drivers to make appropriate decisions and trade-offs.

LLMs can operate at real-time speeds and scale effortlessly across high user volumes. LLM model choice directly affects cloud cost and performance, and that there are a range of different sized models that can be leveraged. While larger models are indeed powerful, they come with higher operational costs and cloud infrastructure requirements that aren’t always necessary for every use case. Smaller, task-specific models can deliver comparable performance without the cost and latency overheads of a massive model.

A retail client originally intended to use GPT-4 to process over 100,000 customer queries daily. After testing, we demonstrated that a smaller, more efficient model could provide comparable performance with reduced latency and lower cloud costs, achieving the desired speed and efficiency without overextending their budget.

Cloud cost optimisation health check

Cloud cost optimisation is essential for balancing innovation with financial sustainability. This practice also frees up funds to address tech debt and empower teams to work within sustainable budgets. Use the following questions to assess and optimise your organisation’s cloud cost management practices.

• Is FinOps embedded as a core practice? Are you utilising FinOps to manage cloud expenses collaboratively across finance, operations, and engineering? Are high-cost or high-impact areas prioritised for optimisation?
• Is precise cost allocation enabled through granular cost controls? Do you use tools and practices to assign cloud expenses directly to teams or business units, promoting accountability and enabling focused cost reductions?
• Are cost drivers beyond major services identified? Are services like Databricks, Managed Kafka, and DynamoDB regularly reviewed for potential cost savings beyond EC2 and RDS?
• Are cloud costs aligned with business value? Do you verify that each service’s cost reflects its business contribution, ensuring critical services justify their expenses and non-essential services are optimised?
• Is right-sizing and auto-scaling implemented to prevent overprovisioning? Are cloud resources scaled based on actual demand, especially in areas with fluctuating requirements, to avoid unnecessary expenses?
• Are data lifecycle policies used to optimise storage tiers? Do you categorise data by access needs, archiving inactive data in lower-cost storage and maintaining high-performance storage only for active data?
• Is precise cost allocation enabled through FinOps and granular attribution? Are costs assigned directly to business units to foster accountability and aid in identifying cost-reduction opportunities?
• Are cloud architectures regularly reviewed for cloud-native optimisation? Do you conduct architectural reviews to ensure applications are leveraging cloud-native solutions, like containerisation and serverless functions, to reduce dependency on over-provisioned resources?
• Are the cost drivers of generative AI applications understood and managed? Do you assess the operational and infrastructure costs of deploying generative AI, such as model size and usage requirements? Are you choosing appropriately sized models to balance performance needs with cost?
• Is there a culture of cloud efficiency supported by accountability and training? Are cost-awareness practices embedded in development and operational teams? Do you employ methods like gamification or designated “cost champions” to motivate proactive cost management?