Technology · AWS
A scaling UK technology business
The client’s AWS estate had grown with the business, and the bill had grown faster, with reservation coverage and instance sizing that no longer matched actual demand. It’s the standard shape of cloud cost in a scaling company: infrastructure provisioned for one stage of growth, quietly carried into the next.
The problem
AWS will discount compute heavily, up to around 72% against on-demand rates through Savings Plans and Standard Reserved Instances, but only in exchange for commitment, and commitment is only cheap when it matches reality. Commit to the wrong instances and you pay for capacity you no longer use; under-commit and you pay on-demand premiums for capacity you predictably need. Both failure modes are invisible on a monthly invoice, which only shows the total.
That’s why coverage decays in growing companies. The engineers who chose the original instance sizes have moved on to new services. The commitments bought two years ago expire quietly, or renew on autopilot for shapes the platform has outgrown. Finance can see the number going up but not why; engineering could find out why, but has features to ship. Industry surveys consistently find that around 29% of cloud spend is wasted, and misaligned commitments and oversized instances are two of the biggest contributors.
What we did
Rather than working around the client’s team, Cloudreason worked through it. We ran the reservation and rightsizing programme alongside their finance and engineering functions rather than on their behalf: coverage and utilisation reviewed with finance, sizing decisions worked through with the engineers who own the workloads.
Rightsizing came first, because the order matters. Committing to an unoptimised estate locks its waste in for one to three years; resize first, then commit to the estate that’s actually worth keeping. From there the programme became a rhythm rather than a project: coverage tracked against a target, expiring commitments dealt with ahead of time rather than after, new workloads sized against evidence rather than habit.
Knowledge transfer was an explicit objective, not a side effect. Finance learned where to look: coverage, utilisation, commitment expiry, and the difference between a good rate and a good decision. Engineering learned how to act: which instances to resize, when to commit, and what a change would do to the bill before making it. The engagement is supported by continuous cost visibility tooling, so both teams can see the position between reviews rather than waiting for the next one.
The results
The work has delivered £180,000 in savings over two years.
The more durable result is the capability. Finance and engineering now share a common view of cloud cost and a common language for acting on it. FinOps practitioners consistently report that the discipline only sticks when it lives inside the organisation rather than beside it, and that is the outcome this engagement was designed around. The consultant who makes themselves unnecessary has done the job properly.
What made it work
The client's finance and engineering teams now share a common view of cloud cost. That's the outcome that keeps saving money after the consultant leaves.
Why this client isn't named
This is a real engagement with real numbers. We don't name clients on our website because we'd rather they weren't fielding sales calls for appearing here. If you'd like more detail, ask us and, where the client is happy, we'll arrange a direct conversation or reference.