How The Best Teams Actually Run AWS

A few years ago, KineticSkunk was called in to help a company after one of its cloud environments had failed. While everything looked fine on paper, recovery took two weeks. Two stressful, hopeful weeks.

We discovered that the business had backups but there was an assumption that stored data meant the environment could be restored easily. In reality, the restore path had never been properly tested against what the business actually needed.

That experience exposed a much bigger truth about cloud operations: getting an AWS environment live is only the beginning. As the business grows, workloads change. And as new services are added, yesterday’s well-designed environment can quietly drift away from the one you thought you were running. Success, in our experience, can often be a double-edged sword.

For AWS, operational excellence is an ongoing discipline. Its Well-Architected Framework (our corporate Bible) says organisations should be able to operate workloads at scale and evolve them over time. Continuous improvement and feedback loops are a must if you’re going to keep operations aligned with changing priorities.

We believe go-live is where cloud implementation ends but cloud operations begin.

The cloud you designed is not necessarily the cloud you are running

AWS gives organisations extraordinary freedom. One of the reasons businesses move there in the first place. Implemented well, it is a boon to your bottom line. But it also means infrastructure must evolve at roughly the same speed as the people using it to keep your business booming.

Someone launches a product and needs more compute. A new customer requires different controls. Engineering experiments with a service that solves an irritating performance problem. Traffic doubles and yesterday’s pleasantly oversized infrastructure suddenly looks rather sensible. Nice problem to have? Not if you’re having it.

Each decision can lead to an explosion of activity which, taken together over time, can create an AWS environment that has drifted quite far from its original design. But don’t get us wrong. This is not necessarily bad engineering. Quite often it is actually a good thing because it’s the technological footprint of a growing business. (Our favourite partners, incidentally.) The problem starts when the operating practices around the environment do not evolve with it.

Cloud drift rarely kicks the door down

The interesting thing about operational drift is how polite it can be. No single action causes an immediate incident. We think of it as the distance or difference between what the organisation believes about its AWS environment and what is actually happening inside it.

The Uptime Institute’s 2026 outage analysis found that around 10% of organisations said their most recent outage had a serious or severe impact. One in five reported costs exceeding $1m. It also found that failure to follow established procedures remains the leading driver of outages involving human error.

Skunk Tip: Good operations involves much more than having good technology. Ownership and operational process need to keep pace with technical reality.

Go-live was the beginning

A successful AWS migration is worth celebrating. But the value of that environment will ultimately be determined by its success.

Good cloud operations means paying attention to the slow accumulation of complexity and revisiting yesterday’s assumptions and continuously improving the environment.

That is also where an AWS partner can become more valuable: providing continuity around the environment and helping internal teams keep the platform aligned with what the organisation is becoming.

Because getting onto AWS is a project. Running it brilliantly is an art.