Worldwide cloud spending is on the rise and shows no sign of abating. Goldman Sachs forecasts cloud computing sales to hit $2 trillion by 2030. Generative AI will account for 10-15 percent of that spending, with a price tag of $1.1 trillion for IaaS and PaaS. At the same time, a significant amount of cloud spending is wasted—as much as 21-50 percent, according to a survey by Stacklet.

How can cloud-driven companies, particularly those running or planning AI projects, plan for profitability? By using Cloud Unit Economics (CUE), a holistic model for forecasting and measuring cloud spending. Let’s dig into how this model can reduce waste and increase profit.

Start with unit cost metrics

CUE starts with unit cost metrics, which measure how much you spend to serve your customers over time. What is the formula for calculating this? Divide your cloud costs by a demand driver of your choice. 

For example, let’s imagine an online retailer, Acme Apparel. Each year, it spends $300 million on the cloud and receives two billion orders, meaning $300 million divided by 2 billion comes to $0.15 per order. Companies with such a high volume like a larger number to work with; in this case, it would be cost per 1,000 orders. That comes to a unit cost metric of $150 per 1,000 orders.

Unit cost metrics underscore the counterintuitive but no less true reality that higher cloud cost doesn’t automatically equal lower cloud efficiency. If your cloud infrastructure is efficient, your cloud expenses may rise even as your unit costs decrease. That means you’re scaling efficiently and growing your business.

Margin downgrades and cost per customer

Unit cost metrics are great for that 1,000-foot view. It’s important to understand how these unit cost metrics are shifting and adapting, and then you need to figure out why.

Margin downgrades reveal the areas of your business most negatively impacting your profit margins. For example, are there specific customers or products that are dragging your margin down.

For a B2B SaaS company, cost per customer yields insightful information. For this metric, allocate your cloud costs to the customers driving them. Many companies simply divide cloud costs by the number of customers, but that only provides an average cost per customer — and not all customers use or need the same capacity. Average cost is not all that helpful in this context. Similarly, using a weighted average against a metric like revenue may also be misleading. This assumes that your largest customers cost the most, which may or may not be the case.

Companies are often shocked when they calculate their true cost per customer. By accurately allocating 100 percent of your cloud costs to the customers driving them and comparing those costs to each customer’s contracted revenue, you can identify your most and least profitable customers and gain clarity about what needs to change.

Once you have the list of customers who are negatively impacting your margins, you’ll want to discover why this is the case and what can be done. Perhaps your discounts were too aggressive, or some customers are using a specific product or feature in a way that was not planned.

Cost per customer might not be the best metric for you, though. You can use cost per feature, cost per microservice, cost per product, cost per engineering team, or any other business metric you find useful. Or you can combine some or all of these to get granular insights into what is affecting the efficiency of your unit cost metric.

The goal: Cloud Efficiency Rate

The apex of Cloud Unit Economics is Cloud Efficiency Rate (CER), a measure comparing cloud spend to revenue. This provides a holistic view of cloud efficiency, compared to companies similar to yours. The formula is simple: subtract cloud costs from your revenue, then divide that number by your revenue, and then multiply by 100.

This number reveals how much revenue you hand back to your cloud providers. Let’s use the Acme Apparel example again. Its annual revenue is $1 billion, and its cloud costs are $200 million, which comes to $800 million divided by $1 billion using the above-mentioned formula. Multiply that number by 100 to get the CER: 80, or 80 percent. This means $0.20 of every dollar earned goes to the cloud provider.

Let’s imagine Acme gets serious about cost savings and discovers many opportunities for greater efficiency. They lower cloud costs by 30 percent while also increasing their revenue by 20 percent. Here’s the math:

($1.2B - $140M)

______________ x 100 = 88.33 percent

88.33 percent is not quite an elite CER, but it’s headed in the right direction – and it’s only a few percentage points away from being in the top quartile (which starts at 92 percent).

Maximizing efficiency

Companies don’t seem to be measuring their cloud costs accurately or with deep and granular visibility, which leads to cloud inefficiency. This is not surprising, given that 61 percent of companies that participated in a recent survey conducted by CloudZero said they had no formalized cloud cost management program.

With AI initiatives ramping up, companies reliant on cloud computing will need even more capacity. That means they’ll need a CUE dashboard to help them predict cloud costs as they scale, measure customer profitability, and price or discount without affecting profit margins. This will be key for accurately measuring unit cost metrics, margin downgrades, and Cloud Efficiency Rate.