It’s a common question from finance, executives, engineering, and DevOps teams. How can we reduce and optimize our AWS cost? After all, at the end of the day, your AWS spend can be a major monthly cost which can be really challenging to become or stay profitable.

So when companies approach me about cutting an AWS bill, I like to work with them by taking a step back and put cloud cost optimizing in two buckets: there are rate optimizations and usage optimizations.

Rate Optimization: This is where you look at the commitments for the underlying resources, enterprise commitments, credit programs. Examples include reserved instances, savings plans, PPAs, EDPs, spot instances, reserved capacity, and other commitments that are on the billing layer. The nice thing about this is while it’s important to loop in and check with your engineering team, no engineering time is needed change anything on the infrastructure. There are some reseller options where they can help do this for you.

Usage Optimizations: This is where you are actually making big changes to your infrastructure. Turning off or deleting unused instances, rightsizing instances, or re-platforming from EC2 to Serverless. These tasks requires engineering resources.

For example, a fast-growing SaaS company realized they were running dozens of dev and staging EC2 instances 24/7, even though no one used them after work hours. By implementing an automated scheduler to shut them down during nights and weekends, they saved over $20,000 annually. That’s a classic usage optimization, identifying what’s running, when, and if it really needs to be.

On the rate optimization side, another company had over $1M in annual EC2 spend but hadn’t purchased any Savings Plans. After analyzing their usage patterns, they committed to a mix of one- and three-year Compute Savings Plans and immediately reduced their EC2 costs by over 30%, without touching a single line of code.

Both types of optimization matter, but usage optimizations tend to have longer lead times, as they require planning, testing, and engineering involvement. This includes tasks like rightsizing EC2 instances based on CPU and memory metrics, or moving from RDS to Aurora Serverless to reduce idle database costs. These changes can unlock major long-term savings, but they often need buy-in and prioritization within engineering roadmaps. And one thing I very much value is engineering time.

Rate optimizations, on the other hand, are often lower-hanging fruit. If you haven’t touched your billing commitments in the past 6–12 months, there’s a good chance you’re overspending. Compute Savings Plans, Reserved Instances, and Enterprise Discount Programs (EDPs) can all drive down your effective rate—sometimes by 30–70%, with minimal operational lift. Some organizations also work with resellers or FinOps consultants who can structure and manage these commitments for them.

Obviously the best way to reduce your cloud cost is to do both of these types of optimizations. You should do rate AND usage optimizations. But sometime, your team is heads down building a new product or tool, or your unsure if you’ll stay with AWS, or you just don’t have the time. Here are some of the best practices for rate AND usage optimizations.

Step One: Take an Assessment / Survey What’s the Situation

The first most critical step is understanding where are your costs on AWS coming from? Are the costs from a specific EC2 or RDS instance, are they from S3 or CloudFront, or are they from many different services.

Let’s say you look at Cost Explorer, and find the following services for the month before:

1/ AWS EC2 - $48,400

2/ AWS RDS - $23,300

3/ AWS EC2-Other - $4,950

4/ AWS S3 - $200

5/ AWS CloudFront - $190

6/ AWS OpenSearch - $100

7/ AWS Redshift - $100

8/ AWS Lambda - $100

By doing this, you are instantly figuring out which costs you should be focusing on. EC2, RDS, and EC2-other! The rest of the costs don’t matter. Next, look at Cost Explorer historically, have these costs always been this high, are perhaps slowly growing, or suddenly are 5 times higher than the month before. And make sure you are working / collaborating with the engineering or DevOps team so that if they just moved some resources from GCP to AWS, then perhaps that’s why the cost went up 5 times. Once you are able to get a good understanding of the costs (and perhaps some of the context behind what’s behind the cost, it’s now time to investigate.

Step Three: Take Action

Now we have successfully identified which instances are costing us the most. Next, we can look at the daily view over the past few months. Perhaps it’s the case that the g5.xlarge and g6e.48xlarge have been running for the past few months and haven’t seen an increase in price. It may turn out that the p4de.24xlarge has been growing by 25% every week, and that’s impacting the price.

Therefore, we’d go look at the instance to understand how it’s being used, figure out if autoscaling is activating, and see if it could be rightsized (maybe instead of a 24xlarge, go down to a 12xlarge, etc.) Here’s a helpful video to rightsize instances:

The example we just went through ended with right sizing being a great way to reduce your cost, but there are other types of optimizing. For example, the above is an example of usage optimizing, but you can also optimize on the rate level. Meaning, let’s say you are running 10 computers, there are two ways you can save money.

You can use less computers (maybe 8), and that’s an example of usage optimization, or you can do rate optimization, where you commit to using 5 of the computers for 18 months and in return you get a 30% discount. If it’s been many months of using that p4de.24xlarge, maybe its the right time to consider buying a reserved instance (a 1 or 3 year contract with AWS promising to use it during that time, and if not using it, you are still paying the cost associated with that.

So in this case, once you find the optimized usage amount, I would next recommend turning to rate discounts. When thinking about how to excel in rate optimizing and usage optimizations, this FinOps Foundation video may be helpful:

Another Tactic: Work with an AWS Partner/Reseller

There are many AWS Partners that are helpful in reducing your cloud costs. Ask you AWS Account Manager if they recommend any in particular, or take a look here.

Some things to keep in mind about a great cloud cost optimization partner:

1/ They are 100% free. The best partners help cost optimizing companies while also being completely free. Make sure your partner is free, because there are high quality free partners.

2/ Make sure they can get high coverage for on-demand spend. Usually aim for above 75% to be at the best place for optimization and flexibility.

3/ Ask them to share customer references or case studies, and ask if they would recommend going with Pump.

4/ If they buy you reserved instances or savings plans, make sure that they are being 100% utilized by you. See more information here.

Based on these things, you should have a good way to find a great AWS Cloud Cost Optimization Partner.

Ultimately, combining both rate and usage optimizations yields the most impactful and sustainable cost reductions. Also, reading these blog articles might help you:

Yaddle Out