Internet Services

What makes internet services different

The environment is usually the accumulated result of shipping quickly, which is the correct thing to have done. Nobody set out to build what’s there now; it grew a decision at a time, and each decision was locally reasonable.

The bill is normally the first symptom anyone reports, but it’s rarely the actual problem. Cost is a lagging indicator of architecture.

How we work here

We start by making the spend explainable — tagging and allocation, so the numbers map to teams and products. That usually surfaces the architectural question the cost was hiding, and it means later reductions can be argued about with evidence rather than feelings.

What we’ve done in this sector

In prior roles, our people worked inside SaaS businesses, one of them carrying an eight-figure annual AWS bill.

That environment went from a single reseller-managed AWS account to a direct Amazon relationship across more than 100 accounts, with governance boundaries drawn so service teams could move independently. Reserved instance purchasing cut EC2 and RDS spend by 40%; policy-based controls cut non-production spend by 65%. On the delivery side, the core product moved to a Twelve-Factor microservice architecture and the release cadence went from monthly to daily, along with server deployment times falling from hours to minutes once infrastructure was under Terraform.

All of it under SOC 2, ISO 27001, and FedRAMP obligations at the same time.