AWS cost optimization that actually gets implemented.

What we optimize

We cover the usual compute, container, storage and networking services, and many more. On compute and containers that means EC2, Lambda, EKS, ECS, Fargate and EMR, plus the managed services that run on EC2 under the hood: RDS, ElastiCache and OpenSearch.

For each of these we apply the strategy that fits the workload: instance type selection, right-sizing, autoscaling and elasticity, adopting Spot instances, migrating to Graviton, and buying Savings Plans and Reserved Instances sized to what you'll actually run.

On storage we pick the right S3 storage class, add lifecycle policies, and right-size EBS volumes and their provisioned IOPS and throughput, including the EBS volumes behind RDS. On networking we cut inter-AZ traffic, add VPC endpoints and PrivateLink, and improve the CloudFront cache ratio. Storage, logging and networking are worth calling out because no Savings Plan or Reserved Instance ever covers them: they get cheaper only when someone optimizes them.

Much of this isn't only about cost. Teams routinely see better performance after the same exercise, most visibly when adopting Graviton, using VPC endpoints and improving the cache ratio.

What it costs

Most teams start with one scoped project: an agreed set of implementations for a flat fee, with the expected savings named before we begin. If it works, it becomes an ongoing monthly retainer.

We work three ways: a scoped project for a flat fee, an ongoing monthly retainer, an agreed block of engineering hours priced case by case against your scope and what you need, or a share of the savings, typically 10% to 30% sized to the footprint under our scope and billed per change for its first 12 months. On the savings-share model, a change that saves nothing isn't billed. All three can go through AWS Marketplace, so the money comes out of the cloud budget you already have. On any model, savings are measured automatically from your real before-and-after billing data, per change, so what you pay always ties back to the actual bill.

It typically costs less than a FinOps SaaS subscription, and on the savings-share model we'll agree a cap if you want one.

Where the savings really come from

Commitments are the easy win, an afternoon of work, but they only ever cover part of the bill. Most of what we find is small: a few hundred dollars a month each, low-risk, with no migration project. Individually, none of them justify pulling an engineer off the roadmap for half a day to research, execute and verify, which is exactly why they never get done.

The arithmetic works for us, because we automated it.

Every change is prepared with help from tooling we built and battle-tested over many engagements, then still ships as a pull request through your own review. What costs your engineer half a day costs us half an hour, so the small wins finally get captured. One piece of that tooling, our EBS Optimizer, is available on its own for teams who just want their EBS volumes right-sized.