<img height="1" width="1" style="display:none;" alt="LinkedIn" src="https://px.ads.linkedin.com/collect/?pid=1310905&amp;fmt=gif">

Discover

Explore CloudZero


Overview Icon
Overview

Discover the power of cloud cost intelligence.

Services Icon
Services

Give engineering a cloud cost coach.

Pricing Icon
Pricing

Learn more about CloudZero's pricing.

Demo Icon
Demo

Request a demo to see CloudZero in action.

About Icon
About

Learn more about CloudZero and who we are.

Connect With Us

Got questions? We have answers.


Questions Icon

Speak with our Cloud Cost Analysts and get the answers you need.

Get in touch arrow-right

Featured

How SeatGeek Decoded Its AWS Bill and Measured Cost Per Customer

Read customer story arrow-right
User Icon

By Role


Engineering

Enable engineering to make cost-aware development decisions.

Finance

Give finance the context they need to make informed decisions.

FinOps

Decentralize cloud cost and mature your FinOps program.

Use Case Icon

By Use Case


Provider Icon

By Provider


Amazon Web Services (AWS)

Measure, monitor, and optimize cloud spend on AWS.

Snowflake

Combine cloud cost intelligence from AWS and Snowflake.

Resources Icon

Learn


Blog

Discover the best cloud cost optimization content in the industry.

Content Library

Browse helpful webinars, ebooks, and other useful resources.

Case Studies

Learn how we’ve helped happy customers like SeatGeek, Drift, Remitly, and more.

Featured

5 Tactical Ways To Align Engineering And Finance On Cloud Spend

Read blog post arrow-right

The New Age of Cloud Cost Optimization: 8 Best Practices

Can’t reign in cloud costs? These practices produce better results than traditional cloud cost optimization methods.

Is your current cloud cost tool giving you the cost intelligence you need?  Most tools are manual, clunky, and inexact. Discover how CloudZero takes a new  approach to organizing your cloud spend.Click here to learn more.

As companies are increasingly built in the cloud, cost optimization has become a major issue for many engineering teams. While cloud providers like AWS offer flexibility and easy scaling, cloud costs can be opaque and difficult to track.

As a result, many companies that rely on the cloud are adopting cloud cost optimization strategies to understand and manage the charges associated with their cloud-based technology, and maximize cloud usage and efficiency. To support this push, a variety of cloud cost management tools, solutions, and software have emerged taking different approaches to cloud cost optimization.

In our view, cloud cost optimization is not just about reducing costs — it’s about connecting costs to business goals. An increase in cost is not necessarily a problem if it’s connected to a corresponding increase in revenue. One of the most important goals in cloud cost optimization is ensuring that costs map to productive and profitable activity. Switching to this proactive approach requires the availability of meaningful data — or what we refer to as cloud cost intelligence.

In this article, we’ll examine best practices for cloud cost optimization done right, and introduce how cloud cost intelligence can help you gain a strategic advantage.

8 Best Practices to Optimize Cloud Costs

The best practices that follow will help you establish a cloud cost optimization strategy that correlates costs to business strategies.

1. Set up your account for monitoring.

If you haven’t already, set up a master AWS organization payer account. As you create new account members, ensure their cost data rolls up into that master account. Then there will be a single source of truth for the company’s total cloud spend. Separate accounts make it very hard to track cloud costs down the road.

Next, work to capture context. There is a lot of telemetry coming out of AWS, from sources like CloudWatch, CloudTrail, and VPC flow logs. These all provide context as to what’s happening in the system. Your team needs to be sure they’re collecting this context in order to be able to tie it to billing.

Finally, start tracking your cost history. Enabling cost and usage reporting allows review of past spending, and helps create context for identifying anomalous costs.

2. Align your internal budgeting and escalation processes with business goals.

A major factor in controlling costs is making sure everyone understands their budgets and goals for each project. Instead of picking an arbitrary number, engineering leaders should have conversations with executives and product leadership to understand cost requirements. Requirements should be based on how the products or features will be packaged and delivered (e.g., Will they be a part of a free trial or enterprise plan?).

These requirements will need to be referenced as tradeoffs throughout planning and development alongside other requirements, such as speed and resiliency.

3. Make cost a first-class metric.

You can do this by promoting the idea of cloud cost within the organization and making it easy for everyone on the development team to see and understand actual costs. Keep cost top-of-mind as engineering decisions are made. To make cost data easy to understand, make sure:

  • It is current.
  • It has context.
  • It can be measured.
  • There is a clear definition of good and bad.

There are several key metrics that pertain to cloud cost:

  • Unit cost: Understand what your unit cost is — is it cost per API call, or cost per report? Whatever the unit, everyone should understand what factors affect cost and how unit cost impacts your bottom line. This metric helps inform decisions about how to spend resources in the cloud and how to charge customers, and will help you understand the ROI on your cloud investment. Spending a lot in the cloud isn’t a problem if the return on unit cost is strong.
  • Idle cost: Determine the baseline cost of your system with zero customer load. This is a general measure of your efficiency. Knowing your idle cost can help you determine whether an architecture change is worth the effort in terms of the savings you can realize.
  • Shared infrastructure: Shared systems can offer cost savings or engineering efficiencies, but make it challenging to split costs across multiple teams. Large organizations should consider how they will charge back or account for shared costs.
  • Cost/load efficiency curve: Calculate your cost/load curve. Do unit costs grow at the same rate as your customer base, or do they grow exponentially as you add customers? If you have an exponential growth curve, you may reach a crossover point where the system is no longer profitable. It’s important to recognize this before it occurs, so you can think through and weigh the solutions for addressing it.
  • Innovation/cost ratio: Determine the ratio of R&D costs to production operations costs. Although there is no expectation of revenue for products in the R&D phase, those products will eventually go to market. If cost has not been considered during R&D, moving to production can be challenging.
  • Cost optimization through the software development lifecycle: Too often, cost only becomes a consideration after a product has been built and launched. Ideally, cost optimization should be a concern throughout the software development lifecycle. Every engineering decision has an associated cost; supplying the right data to the right people at the right time can help them make decisions to optimize costs.

Looking for a cloud cost optimization tool that automatically maps costs to the activities that generated them? Schedule a CloudZero demo to see how it works.

5. Define other metrics important to business goals.

It is crucial that everyone in the business whose work impacts cost understands the fundamentals of the business and uses this information to drive decision-making processes. DevOps should strive to build applications, features, and services to support both customers and revenue. They should also, at the VP level and sometimes below, understand capital flow with respect to engineering investment.

Everyone should understand the most important goals of the organization at any given point in time. For startups, the goal may be customer growth above all else. For more mature organizations, the goal may be to increase margins. When DevOps understands what these goals look like, it’s possible to make day-to-day and high-level decisions that center around what is best for the business alongside what is valuable to customers.

Metric

Questions To Ask / Questions It Can Answer

cost per feature

  • How much will it cost to build this feature?
  • How much incremental revenue will this feature bring in?
  • Is this feature worth building right now? (Or is there another one that would map to business goals better right now?)
  • How much are legacy decisions costing us?
    • If we were to tackle tech debt, which projects would provide the most “bang for buck” in the short term? Long term?
  • What parts of application affect costs in a disproportionate way?
    • Do they justify themselves via revenue?

cost per customer and segment

  • How much does a new customer cost us in terms of cloud storage, compute, etc.?
  • How much is each new customer worth in terms of revenue?
  • Are we making or losing money on certain types or sizes of customer?
  • Is there a better way to price offerings to account for costs?
  • Are segments such as enterprise customers higher or lower margin than other segments?

cost per app

(or cost of platform, etc.)

  • How much does it cost to run our app daily? Monthly? Yearly?
  • How does that cost change with customer acquisition?
  • Are we making money on the app?
  • What are our margins?

cost per team

  • How much is each team spending in the cloud?
  • Are there opportunities for cost sharing or economies of scale?

revenue

  • How much is the company making on the platform/app(s)?
  • What is the projected annual revenue?

cloud cost

  • What is the annual cost of running our entire cloud?
  • How is that expected to grow or shrink over time?

cost per unit*


*the definition of “unit” varies based on what you are selling and how; see Unit Cost section above

  • What is our cost per unit?
  • What is our revenue per unit?
  • What are the margins?
  • Are there opportunities to reduce cost per unit?
  • Is our business model mapped well to our cost per unit?

time to market

  • How long does it take to get products / features to market?
  • How much does that cost in terms of man-hours?
  • Are there ways to decrease time to market? (And thus decrease the overall cost of delivery?)

cost per cloud service

  • How much are we spending on:
    • Storage?
    • Compute?
    • Databases?
    • Other cloud services?
  • When cost spikes arise, which services are affected? (e.g., logging is a common culprit)

cost of R&D

  • How much are we spending on R&D?
    • Man hours
    • Technology costs
  • What is the potential value of outcomes (products, services, etc.)?
  • What types of R&D efforts are worth our time?

cost deviations

  • When a cost spike occurs, what is the cause?
  • Is the increased cost balanced out by increased revenue?
  • If not, how can we put a stop to the spike?

6. Get the right data to the right people at the right time.

More data is not always better; in some instances, too much data is part of the problem. For this reason, it’s worth understanding what types of data are most useful to which team members and find ways to decrease the noise of extraneous data.

For example, engineers need to be able to compare and contrast. With baseline data and the ability to look backwards, DevOps can understand whether something is broken or where an anomaly has arisen. DevOps organizations also need to be able to slice and dice data by resource and by team, as well as by feature and service. With this type of granular data, they can fix problems faster.

Finance, on the other hand, often cares more about forward-facing projections. For example, they may want to know: “How much will our cloud bill increase if we add six new customers this month?” or “How much can we expect to spend on the cloud before the fiscal year is over?” Finance will care most about return on investment—in other words, how much revenue are we getting from our investment in the cloud?

Ultimately, engineering and finance are looking at the same data, but they need to see it in different formats and be able to organize it in different ways in order to answer the questions that are most important to them in their roles. When you think about surfacing data, you also need to make sure that data is relevant — that you are able to give the right people the right data at the right time.

7. Use data to optimize cloud costs at each stage of the software development lifecycle.

Opportunities exist at each stage of the software development lifecycle to optimize cloud spend:

  • Planning stage: Teams should be able to justify the budget they need and use cost data to inform product roadmap and technical debt-related decisions. This enables them to reduce unexpected spend and quickly adjust the budget when necessary.
  • Deployment and operation: Teams should be able to quickly identify unpredicted spend and adjust.
  • Design and build stage: Teams should have the data they need at hand to make cost-effective architecture decisions. They should be able to report on planned spend and understand their cost of goods sold (the unit cost).
  • Monitoring stage: It should be possible to reassess cost by team, product, or feature and report on operational expenditures and ROI, segmented by business initiative.

When teams have access to the right data at the right time, they are more able to “pull the right levers” — in other words, make timely changes that positively impact both product quality and the bottom line.

8. Weaponize cloud costs for business strategy.

Most organizations default to charging customers based on the metrics they already have access to. In other words, they “back themselves into” their pricing strategy. Instead of picking the metrics that are best for the business, they pick the ones they can already track. This makes sense on the surface, but it often becomes an issue when pricing strategies are not tied directly to cloud costs.

For example, a business that offers code scanning might charge by megabyte of code scanned, but maybe charging by line of code would make more sense in terms of business outcomes. Customer growth may lead to nonlinear cost increases in the cloud, and leave the business struggling to make up for the margins.

All features of a product should have a dollar cost associated with them. If DevOps starts by setting up the application to measure the metric mentioned in this guide from the very beginning, then it’s possible to develop a very keen sense of how much individual features and transactions cost the business. This gives product teams flexibility to creatively price offerings, since they will have a very clear and direct sense of how much it costs to run and scale.

Ultimately, this level of visibility enables companies to create an ideal business model and to unlock new opportunities. If you use cloud cost transparency to your advantage, you can turn price into a competitive weapon that can help you outpace competitors.

A Modern Cost Optimization Strategy: Cloud Cost Intelligence

The above best practices for cloud cost optimization all require a lot of data — data that is linked to specific teams and cloud features.

While many cloud cost optimization tools can help you reduce cloud spend to some degree, few have the ability to provide the detailed data needed to reasonably administer these best practices. Most cloud cost optimization software products essentially look backwards, and lack the ability to map costs directly to the activities that generate them. They may reveal when a cost spiked yesterday (after it has already resulted in an unanticipated expense), for example, but typically, they will not be able to pinpoint the precise event or feature that produced the spike. Instead, it’s up to your developers to pore through a lengthy report to determine what caused the spike — which results in additional costs. The cost anomaly may continue until the source is detected, and developers will waste time looking for the cause instead of focusing on more valuable work.

When you can automatically connect costs to the specific teams, events, or features that generate them, you will have moved beyond traditional cloud cost optimization to cloud cost intelligence.

Cloud cost intelligence tells you where you’re spending your money and (more importantly) what that means in the context of your business. It continuously delivers the data stakeholders need to detect and fix anomalies — and design and build cost-efficient products — automatically correlated to the activity generating the cost. Cost data is connected to individual features, events, and teams, so there is no need to search for the cause of unexpected cost spikes. In short, cloud cost intelligence represents the ultimate in transparency, giving finance, developers, engineers, and any other relevant parties the information they need, when they need it, to control cloud costs.

You can achieve cloud cost intelligence with the help of a cloud cost optimization solution like CloudZero. CloudZero is the only cloud cost intelligence platform that puts engineering in control. It aligns your cloud costs to teams, customers, unit cost KPIs, product features, and more — so you can stop guessing and make cost-informed decisions.

CloudZero Cloud Cost Management

To ensure that the right people on your team get the right data at the right time, CloudZero:

  • Pulls in both billing and resource data from across your AWS account to automatically group costs and surface insights.
  • Aligns cost to business metrics, without manual tagging effort.
  • Updates continuously, to give you real-time cost data.
  • Sends automatic anomaly reports to the appropriate teams.
  • Runs cost and resource data through a powerful data normalization and machine learning engine to associate costs with the activity or teams that produced them.
  • Gives a clear picture of how your costs are trending and where you need to focus to control nonessential costs.

Want To See CloudZero in Action?

We’d love to give you a tour! If you’d like to see how cloud cost intelligence can boost your organization’s bottom line, sign up for a CloudZero demo today.

STAY IN THE LOOP


Join thousands of engineers who already receive the best AWS and cloud cost intelligence content.