In SaaS companies, engineers are the biggest influencers to cloud costs. They choose the infrastructure, build the products, and produce the code.
Unfortunately, having this power means engineering managers are often asked to predict cloud spend months or even years into the future.
An executive or a head of finance might approach the engineering head and ask how much the company will spend on cloud costs next year, thinking he or she should naturally have the answer.
The problem is that it’s hard to predict the future in any regard, let alone in such a complex situation as a SaaS environment.
Imagine if someone asked you to predict your household spending a decade from now. Consider all the variables that might change your answer. Perhaps you’ll have three more kids, you or your partner could switch careers, or you’ll inherit a huge windfall from a distant relative.
Don’t forget to account for inflation and market changes on top of that!
Extrapolate that to a company’s entire line of products — and remember, SaaS products require dynamic and extremely complex budgets, because costs always depend on a wide number of factors — and you can see why engineering managers can get frustrated by the responsibility of predicting next year’s spend with any degree of accuracy.
Interestingly, however, there is a way to keep all parties happy.
With the right platform, you can learn to predict cloud costs within a reasonable margin of error. This gives executives what they want, provides finance with the information they need to construct their budgets, and relieves the stress engineers feel about taking shots in the dark.
The key is to choose a cloud cost intelligence platform — such as CloudZero — that allows you to break costs down into unit economics. Let’s discuss how unit economics can help you unlock accurate cost forecasting and what you can do to get started.
How Tracking Unit Economics Leads To More Accurate Cloud Cost Predictions
There are a few major drivers of cloud cost changes for SaaS companies, and adding new functionality to your existing lineup, for instance by releasing an entirely new product or adding some features to an existing one, is a prime example.
Perhaps the new product or feature consumes resources your other products don’t use as much, or maybe your new release attracts a different kind of customer than you’re used to serving.
There are any number of variables at play, but regardless, you can count on definite changes to your cost profile any time you ship something new.
If all you know right out of the gate is that you’re releasing a new product or feature and that your costs will probably change, that’s not very useful for making prediction months or years into the future.
In fact, any significant changes to your existing system are almost guaranteed to cause a big change in your costs. And unless you can model your current costs in detail, it’s hard to understand how future changes may impact your overall bill.
But what if you knew from experience exactly what it costs your company to ship a new product? Or, what if you knew how much the services cost to support your new feature because they’re similar to the services used for features of your other products?
If you also knew how much it costs your company to support each customer, you could construct a set of predictive models using this data to account for a range of possible outcomes.
In this example, the relevant unit economics you’d need to calculate include:
- How much it costs for your company to build a new product or feature
- How much the average customer costs your company
- And perhaps the costs per customer broken down across specific products, features, regions, business sizes, or other categories.
The further you can break your costs down, the more powerful your unit economics data will be.
However, your goal isn’t simply to collect this data and file it away. Let’s take a look at how you could take your company’s unit economics data and transform it into a useful predictive model.
Using Unit Economics To Forecast Future Cloud Costs
Let’s say you’ve been tasked with predicting your company’s cloud costs over the next year. If your company tracks its costs in enough detail, all you have to do is ask a few questions and you’ll have everything you need.
You could go to the sales team, for example, and ask about their sales plan for the next year.
They may have different plans for different customer segments. One common example is how sales often targets small businesses and large enterprises very differently, even if both segments end up using the same product at the end of the day.
Next, you can approach the engineering team and ask what products or features they plan to ship within the next year, and what they expect the functionality to look like.
Finally, you could talk to the product team and ask how different types of customers use your company’s products. Maybe large enterprises tend to churn through computing power and storage at a high rate, while small business users go light on compute and storage but expect in-depth and ongoing customer support to help them use your product effectively.
For the sake of the example, let’s say you learn that the sales team wants to double-down on attracting small business customers, so the engineering team plans to build a new feature that will make accessing live, ongoing support more convenient.
It may be helpful to ask the product team how they expect user demographics to change over the next year in light of the planned feature release.
You can now use this information — in combination with your granular cost data — to build a set of predictive models.
Look at the changes the engineering team plans to make. Is their new feature similar to anything your company has released in the past?
If so, you have a fairly good idea of what the new one might cost. And if not, you can do some more digging to uncover what resources the new feature will use, and the costs associated with the expected increase in those resources. Don’t forget to account for related costs such as marketing and ongoing updates.
Next, take the estimates of how the company’s user base will change. If a 20% increase in small business customers is expected, you can look at what it costs to support your small business customers and use that information to calculate your prediction.
If you anticipate a shift from enterprise clients to smaller businesses, you can calculate that as well.
In short, if your company tracks individual costs at a deep-enough level that lets you calculate unit economics, there’s almost no limit to the predictive cost models you can build.
Put enough “what if” scenarios together into these predictive models, and you’ll be able to choose the most financially beneficial path forward for your company.
CloudZero Can Help You Break Costs Down Into Unit Economics
Did you notice the big “if” statement above? Everything hinges on your ability to track costs at a deep and granular level.
You won’t be able to jump into advanced cost tracking overnight and wake up understanding all there is to know about unit economics within your company.
Plus, your cost visibility goals will change over time depending on the venture stage your company is currently in — whether it’s a brand-new startup, a mid-size company, or a thriving enterprise.
One thing all companies have in common is that they need to consistently measure metrics on which to base their unit economics calculations.
After all, it’s hard to determine your costs per customer if you don’t know how to measure how many customers you support and their individual contributions to your cloud costs.
CloudZero takes away all the guesswork and gives you deep visibility into the inner workings of your cloud costs. Track all the metrics that are relevant to your business and see in detail how your costs are affected by factors like customer demographics, new feature releases, and product updates.