Accurate, fair budgets make everyone in the company happy. Engineers love when they can build products that make the company money, executives enjoy seeing nice, wide margins, and finance departments celebrate when everything goes according to plan.
But keeping the budget for cloud computing reasonable can seem like a lofty goal when your company’s cloud spending seems to change with the direction of the wind.
The trick to developing a reasonable cloud budget is understanding how your architecture and customer usage drive your cloud bill. You’ll have to drill down into unit economics as you’ve never done before.
The steps below will walk you through the process but remember: If you get stuck, CloudZero’s expert team is just a few clicks away. We would be happy to help you with your unit economics strategy.
Build Your Budget For Cloud Computing With These 8 Steps
1. Determine your costs for every product
The most successful SaaS companies understand within a fairly sharp margin of accuracy how much each product costs to operate.
You want to break down your costs for every product you offer to see how each product compares to the others.
Perhaps product A costs far more to develop but has the potential to bring in enough revenue to offset those costs, while product B is cheap but has less revenue potential in the long term.
It’s important to understand product-level profitability before you move on any further.
2. Track your product costs by environment
It’s common to have multiple versions of your software running in the cloud. Beyond main Production instances, many companies have staging environments, QA environments, and development environments.
Your list of instances or environments will vary. Still, the idea is to track every dollar associated with a given product in a specific environment, such as:
3. Understand your costs per feature
Similarly, break each product down into its component features and determine the costs for each feature. Remember, the deeper you go and the more granular your view, the more accurate your forecasts will be.
Above is an example of how CloudZero can align cloud spend to cost per feature, product, environment, customer, team, and more.
If possible, break feature costs down by environment — as you did with product costs — to determine how much you’ve spent on research, development, support, and other relevant production stages.
4. Calculate your costs per customer
It often seems logical to focus mainly on the factors directly under your control, such as the costs associated with product and feature development. However, it’s essential to understand how customer usage affects your costs.
Understanding customer usage will allow you to understand what cloud infrastructure growth is good and what growth is worrying. That context will enable you to make better decisions when reviewing your budgets.
5. Begin putting these results into context
Once you break down your costs for every customer, patterns may emerge even before you begin to make future predictions. Ideally, you should be able to break customer costs down into segments, such as enterprise clients and small businesses or free-tier and premium-tier.
You’ve also figured out what it costs to deliver one unit of value to your end customer in some instances. Over time, you can do this for every product, feature, and customer. These are the numbers on which you’ll be basing your forecasts.
If you know how much it costs to deliver product A to customers W and X and product B to customers Y and Z, you’ll be able to make an educated guess about what might happen if you were to make specific changes.
If, for example, you acquired many new customers similar to customer X, you would have a relatively good idea of the impact that might have on your cloud bill. Or, if you have a significant addition of new features in the pipeline, you can evaluate the costs of similar existing features and come up with a prediction on what you might ultimately need to spend on the new features.
If your new feature is still six months out, it’s likely not practical to try and arrive at an exact number. But you can look at the average costs of similar features you’ve developed in the past and find a comfortable ballpark estimate. You might need to adjust as you go along, but you’ll have a solid place to start.
6. Hone your forecasts by creating models
Now it’s time to bring it all together and create some models to forecast costs based on the changes coming down the pipeline for your business.
- What might happen, for example, if you developed three new features?
- What would your costs look like if your customer growth suddenly upturned after these feature releases?
- What if most new customers were similar to customer Y instead of customer X, as you might have originally anticipated?
You have the information to start answering questions like these.
Start small when you first begin. It’s far better to have an accurate forecast of the next month or quarter than to rough guess what things might look like in a year.
7. Improve accuracy
Since forecasting cloud costs and developing a budget for cloud computing are not perfect sciences, you might find that your first few predictions are off the mark.
One common mistake people tend to make is attempting to look into the future before they fully understand the historical data.
It’s tempting to take the past month or two of cost data and extrapolate your findings to predict up to a year’s worth of costs in the future. However, looking back further into the past can help you recognize patterns you might have missed.
Support you continually over- or under-estimate your costs in a certain situation. In that case, you might see these errors indicating that you can adjust your models in one direction.
Perhaps you started off underestimating your future customer growth to be conservative. But after several months of consistently outperforming your predictions, it’s likely safe to start building higher customer growth into your predictive models to improve the accuracy of your forecast.
8. Communicate effectively about variables
Let’s say you’ve developed a few predictive models and arrived at a handful of forecasts for the most likely scenarios over the next quarter. It’s not enough to report those numbers to the relevant team members and move on.
Predicting the future is hard in any situation. But developing accurate predictions when dealing with millions of cloud resources that grow and change according to multiple variables is even more difficult.
Your team should understand that the numbers you’re giving them are forecasts, not hard numbers set in stone. Don’t tell everyone your cloud costs will grow by 20% unless you can also report why you think this is likely and how this final result could be affected by different variables along the way.
Let CloudZero Help You Accurately Predict Cloud Costs
Your predictive model will only be as accurate as the data on which it is based. If you’re developing a model from some back-of-the-napkin math where you divide your total evenly by each end user, you’re forfeiting the most powerful aspects of cloud cost intelligence. At best, your prediction is a rough guess that may or may not bear out over time.
“But wait,” you protest, “how else am I supposed to do it? I don’t have a way to track costs for individual products, let alone for every single customer!”
That’s where CloudZero changes the game.
CloudZero gives you an in-depth look at where every dollar of your spending comes from. You can see costs portioned out by environment, by product, by feature, and by customer as you see fit.
If those elements seem similar to the steps above, that’s by design. We want to give you everything you need to build robust predictive models that can help you plan an accurate budget for your cloud computing needs.