If we’ve learned anything from the constant evolution of cloud computing and its integration into enterprise operations, it’s that generic metrics are rarely good at measuring success. Don’t get me wrong, cloud units are better than no metrics. Most enterprises use no metrics at all, even if they won’t admit it.
Metrics must be aligned with cloud value, otherwise enterprises could get into trouble. Many of those running finops in Global 2000 enterprises are likely to agree with me. Thus, I felt compelled to write this.
What are cloud units?
The concept of cloud units originated within the cloud finops space. Cloud units are designed to tie cloud costs and resources to a standardized unit of business value, such as cost per user, transaction, or workload. The hope was to make cloud economics relatable and actionable for business leaders. On the surface, this sounds like a great idea. After all, in a world where CFOs are left scratching their heads over wildly fluctuating AWS bills, attaching real-world outputs to cloud expenditures offers clarity—or so we’re told.
In actuality, cloud units fall short in most real-world enterprise scenarios. Why? They reduce complex, dynamic cloud ecosystems into overly simplistic metrics that fail to account for the nuanced priorities, goals, and strategic outcomes unique to each business. What starts as a standard measure to align IT and business more effectively often ends up as a distraction—or worse, the wrong tool for the job.
A better method is to measure cloud value with metrics that adapt to your business’s specific needs and evolve with them. For example, an e-commerce platform might measure the cost of completing a single order. This allows leadership to better reconcile cloud infrastructure expenses with operational outcomes.
Where cloud units get it wrong
The one-size-fits-all approach of cloud units might benefit teams just starting their cloud finops journey, but it rarely holds up at scale or when business contexts grow more complex (like every cloud project I worked on in the past seven years).
Let’s call out what’s painfully obvious: Cloud environments are inherently chaotic. Enterprises run thousands of workloads across hundreds of services, often spanning multiple regions or cloud bands. This complexity makes it nearly impossible to reduce an entire cloud ecosystem to a single unit cost. An overly simplistic model often results in more confusion than clarity.
For instance, say your cloud unit represents cost per transaction. If this cost increases, what does that mean? Are inefficiencies driving the uptick or is it a sign of additional investments in scaling infrastructure during a seasonal surge in demand? Cloud units don’t offer that kind of granularity. Instead, they paint with broad strokes, making it easy to misinterpret valuable strategic spending as waste.
Every enterprise operates differently
Some companies focus on customer experience. Others pour resources into creating innovative products. Many have business models that don’t fit a standard mold. Yet cloud units are, by definition, one size fits all. They assume that all workloads should be mapped to the same primary output: the lowest cost per transaction, cost per gigabyte, or cost per instance hour.
For instance, a healthcare organization trying to calculate the cost of securely storing patient records won’t appreciate being lumped into the same framework as a video streaming service measuring costs per viewer. Likewise, an AI startup training machine learning models won’t find value in cost-per-user metrics since their business drivers are entirely different.
What cloud units miss is the individuality of enterprises and their goals. They force businesses to align with the metric instead of aligning the metric with the company. This makes cloud units ill-suited for enterprises that don’t fit neatly into their constraints.
Here’s where cloud units hit an even more significant snag: Value isn’t always financial. Yes, reducing per-unit costs is essential, but what about strategic gains? Enterprises expect outcomes such as greater agility, innovation, or customer satisfaction from cloud investments. Yet, these benefits are hard to quantify using a standardized metric like cloud units.
Consider an enterprise that invests heavily in real-time analytics. The immediate output might show relatively high costs but also long-term competitive advantages like better decision-making or improvements in customer retention. Take an organization that builds cloud-based disaster recovery systems: The value lies in ensuring resilience, not cutting costs per unit. Metrics must be equipped to reflect these intangible benefits so that companies don’t miss the bigger picture.
The case for bespoke cloud metrics
Now that we’ve dismantled the argument for cloud units, let’s talk about what works better: bespoke metrics, which means metrics for each problem domain. This is precisely what most finops pros attempt to avoid, so I suspect I’ll get a lot of pushback for my opinions.
Instead of forcing every workload or department into a generic mold, bespoke metrics reflect your organization’s needs. A media company might track cost per stream, retail operation, and cost per fulfilled order. These metrics are far more relevant than trying to shoehorn cloud spending into a generalized cloud-unit framework.
Custom metrics can balance tangible costs with intangible benefits. For example: time-to-market improvements associated with cloud deployment, customer satisfaction tied to reduced application latency, or revenue growth tied directly to AI-driven innovation projects.
Where cloud units smooth details into one overarching value, bespoke metrics enable deeper inspection. You can track individual workloads, application teams, or geographic regions, identifying areas for optimization without losing sight of broader trends. This precision empowers data-driven decision-making.
A word of warning
I want to be clear: Using metrics to determine cloud value is essential. My concern is that the finops community is oversimplifying something that is complex, and that will lead to an incorrect understanding of cloud value. I want to avoid having companies with great cloud unit metrics watch value draining from their business and have no idea how it is happening. I’m seeing this today, and while it may be just an annoyance, it could become a real problem as finops leads them in the wrong direction.
As an architect, I’m focusing on returning the most value to the business through the proper configuration of technology. This is sometimes more difficult than many understand, and if I’m chasing windmills called cloud units, I can’t make this work. Finops people, time to begin rethinking this one.