McKinsey points to a persistent “on‑premises computing” mindset as the root cause of several recurring mistakes that limit the value of cloud adoption. Key mistakes include: 1. **Treating cloud as a simple lift-and-shift hosting change** Many companies move applications to the cloud as-is to capture quick savings on hosting, storage, and maintenance. While this can reduce some costs on Day One, it often carries over existing technical and operational inefficiencies. As a result, organizations miss out on the cloud’s flexible infrastructure, slower time-to-market improvements, and advanced capabilities. McKinsey notes that **“Year One” benefits can exceed “Day One” benefits by 15% to 25%**, but only if companies plan beyond a basic lift-and-shift. 2. **Clinging to an on-premises, capital-expenditure mindset** On-premises IT is typically financed as capital expenditure (CapEx): you buy and own capacity up front. Cloud, by contrast, is an operational expenditure (OpEx) model where you pay for what you consume. Some companies fail to adjust to this and don’t adopt a dynamic OpEx approach. They don’t precisely measure demand or manage capacity in real time. McKinsey emphasizes that efficient cloud economics now depend on being able to evaluate capacity demand and **incremental or marginal costs at any given moment**, so you pay for capacity only when you actually need it. 3. **Forecasting cloud spend based on history instead of business priorities** When shifting to OpEx, many organizations still rely on historical spending patterns to budget for the future. In the cloud, history is a weaker predictor: usage can change quickly with new products, promotions, or business models. McKinsey notes that this approach often leads to **forecast errors of more than 20%** versus actual cloud spending. Instead, companies should tie cloud budgets directly to business priorities—such as a major Black Friday promotion or a move to a subscription model—because these initiatives significantly change both usage and cost patterns. 4. **Not differentiating workloads by elasticity and demand patterns** Cloud is especially valuable for workloads with highly variable or seasonal demand. McKinsey cites a video-streaming company that measured cloud costs per subscriber and aligned compute needs with demand patterns, achieving **more than 95% accuracy in predicting cloud consumption**. Many companies miss similar savings because they treat all workloads the same. They don’t distinguish between: - Workloads with short-term swings (e.g., traffic spikes during campaigns), and - More stable workloads (e.g., long-term subscriber data storage). Each workload needs to be examined individually to determine whether its elasticity pattern makes cloud more economical. 5. **Weak coordination between cloud architecture and cloud economics** Some organizations overestimate both their cloud usage and the value they will achieve because they plan technology architecture and financial models in isolation. McKinsey notes that while advanced cloud-native enterprises can achieve **resource utilization rates above 60%**, most companies are **below 30%**. To close this gap, businesses need to tightly link their cloud business case with their cloud-architecture transformation. 6. **Assuming everything should move to the cloud** Not all workloads are best suited for public cloud. For a small number of massively scaled, homogeneous workloads—such as certain storage services—custom-designed on-premises infrastructure can sometimes match or even beat cloud economics. Companies with these environments need to be selective about what they move, rather than assuming a full migration is always the right answer. In short, the main issue is treating cloud like a traditional IT purchase instead of reimagining how technology is consumed, financed, and managed. Addressing these mistakes helps companies move beyond basic cost savings and unlock more of the cloud’s strategic value.

Moving to the cloud reshapes your IT cost structure and requires a different financial mindset than traditional on-premises technology. Here are the key shifts for CFOs and finance leaders: 1. **Move from CapEx to dynamic OpEx** On-premises IT is typically a capital expenditure: you buy servers, storage, and data centers up front and depreciate them over time. Cloud turns much of this into operational expenditure: you pay for what you consume, often monthly. McKinsey stresses that the most efficient cloud economics now depend on: - Precisely measuring demand, and - Understanding incremental or marginal costs at any point in time. The goal is to **pay for capacity only when you need it**, instead of funding unused capacity years in advance. 2. **Tie cloud budgets to business priorities, not just historical spend** In a cloud model, past spending is a weaker predictor of future costs. Usage can change quickly with new initiatives. McKinsey notes that relying on historical patterns alone often leads to **forecast errors greater than 20%**. Instead, budgets should be built around business priorities and events, for example: - A major marketing push before Black Friday, - Launching a new digital product, or - Introducing a subscription-based revenue model. Each of these can significantly change cloud usage and cost profiles. 3. **Build a strong FinOps capability** McKinsey highlights the need for a competent **FinOps (Financial Operations)** function that brings finance, technology, and product teams together. The role of FinOps is to: - Help application owners understand the business drivers behind their cloud spend, and - Connect cloud spending to unit economics (e.g., cost per subscriber, cost per transaction). For example, a video-streaming company that tracked cloud costs per subscriber was able to align compute resources with demand and **predict cloud consumption with more than 95% accuracy**. 4. **Segment workloads by elasticity and cost behavior** Not all workloads behave the same way financially. Finance teams should work with technology leaders to: - Identify workloads with highly variable demand (e.g., campaign-driven traffic, seasonal peaks), where cloud elasticity can create meaningful savings. - Separate them from more stable workloads (e.g., long-term data storage), where the economics might be closer between cloud and on-premises. This segmentation helps you decide where to lean into cloud elasticity and where to consider alternative options. 5. **Evaluate when on-premises still makes sense** For a small number of massively scaled, homogeneous workloads—such as certain storage services—McKinsey notes that custom-designed on-premises infrastructure can sometimes deliver economics that are equivalent to or better than public cloud. As a CFO, this means: - Avoiding a blanket “everything to cloud” policy, and - Supporting a hybrid approach where some workloads remain on-premises if the business case is stronger. 6. **Integrate architecture and financial planning** Cloud architecture decisions (e.g., how applications are designed, how resources are provisioned) have a direct impact on utilization and cost. McKinsey observes that while some advanced cloud-native organizations achieve **utilization rates above 60%**, most are **below 30%**. Finance should partner closely with technology teams so that: - The cloud business case is built alongside the architecture roadmap, and - Utilization targets and cost outcomes are tracked together. By rethinking cloud as a flexible, consumption-based utility rather than a fixed asset purchase, CFOs can help their organizations better align technology spending with business demand, improve forecast accuracy, and make more informed decisions about where cloud—and where on-premises—creates the most value.

A full “everything to the cloud” strategy is not always the best fit. McKinsey’s analysis suggests that organizations should be selective and base decisions on workload characteristics, economics, and business priorities. Here’s how to think about it: 1. **Cloud is well-suited for variable and growth-oriented workloads** Cloud tends to be most valuable when: - Demand is highly variable or seasonal, - You need to scale up and down quickly, or - You want faster access to advanced capabilities and shorter time-to-market. McKinsey notes that when companies go beyond a simple lift-and-shift and redesign for cloud, **Year One benefits can exceed Day One benefits by 15% to 25%**, driven by speed and flexibility. 2. **Some large, stable workloads may be more economical on-premises** For a small number of massively scaled, homogeneous workloads—such as certain storage services—custom-designed on-premises infrastructure can sometimes match or outperform cloud economics. In these cases, the scale and uniformity of the workload can justify on-premises investments, especially if usage is predictable and doesn’t require frequent scaling. 3. **Hybrid strategies can balance cost, control, and flexibility** Many organizations find value in a hybrid approach: - Use cloud for workloads with fluctuating demand, experimentation, and rapid innovation. - Keep select, large-scale, stable workloads on-premises when the financial case is stronger. McKinsey emphasizes that businesses should **examine workloads individually** to understand their elasticity patterns and cost implications before deciding where they should run. 4. **Architecture and economics must be planned together** Overestimating cloud usage and value often happens when technology and financial planning are disconnected. McKinsey notes that while some advanced cloud-native organizations achieve **resource utilization above 60%**, most companies are **below 30%**, which weakens the economic case. To avoid this, organizations should: - Align cloud-architecture decisions with a clear business and financial model, and - Continuously measure utilization and cost per unit of business value (e.g., per subscriber, per transaction). 5. **Use FinOps to guide workload placement decisions** A strong FinOps capability can help you: - Compare cloud vs. on-premises economics for each workload, - Understand how business events (like promotions or new products) change demand, and - Decide which workloads should move, stay, or be redesigned. For example, McKinsey highlights a video-streaming company that measured cloud costs per subscriber and aligned compute resources with demand, achieving **over 95% accuracy in predicting cloud consumption**—a level of insight that directly informs where workloads should run. In practice, the most effective strategy is usually not a blanket move but a thoughtful mix: reimagine how each workload is delivered, decide where cloud creates clear value, and recognize where on-premises infrastructure still plays a role in a balanced, cost-conscious technology portfolio.