We Cut AWS Costs 38% in 2 Weeks – Here’s the Exact Process
Client: SaaS company, £2.3M ARR, 8-person team
Starting AWS bill: £8,400/month
Ending AWS bill: £5,200/month
Savings: £3,200/month (£38,400/year)
The Problem
Their AWS costs were climbing faster than revenue. The team knew something was wrong, but nobody had time to audit everything. Every engineer had a different theory about what was wasting money.
Sound familiar?
Week 1: Infrastructure Audit
Day 1-2: Resource Inventory
- Discovered 47 EBS volumes attached to terminated instances (£340/month waste)
- Found 12 unused NAT Gateways across dev/staging environments (£420/month)
- Identified 8 Elastic IPs not attached to running instances (£30/month)
Day 3-4: Compute Analysis
- Production RDS instance running db.r5.2xlarge – average CPU usage: 8%
- Right-sized to db.r5.large = £680/month saved
- EC2 instances: 60% running 24/7 with <10% utilization overnight
- Implemented auto-scaling schedules for non-production = £520/month saved
Day 5: Data Transfer Costs
- Found £1,200/month in cross-AZ data transfer
- Root cause: Load balancer and database in different availability zones
- Fixed architecture = £1,200/month saved
Week 2: Implementation
We didn’t just hand them a report. We implemented every optimization:
- Deleted orphaned resources
- Resized databases with zero downtime
- Configured auto-scaling groups with proper schedules
- Migrated resources to optimize data transfer
- Set up CloudWatch alarms to prevent regression
Results After 30 Days
| Category | Before | After | Savings |
|---|---|---|---|
| Compute (EC2) | £3,200 | £2,150 | £1,050 |
| Database (RDS) | £2,100 | £1,420 | £680 |
| Storage (EBS/S3) | £890 | £550 | £340 |
| Data Transfer | £1,410 | £210 | £1,200 |
| Other | £800 | £870 | -£70 |
| Total | £8,400 | £5,200 | £3,200/month |
The 3 Biggest Mistakes We See
1. Running Production-Sized Resources in Staging
Staging doesn’t need the same capacity as production. We see teams running identical infrastructure 24/7 for environments that get used 20 hours/week.
Fix: Automate staging environment shutdown overnight and weekends. Immediate 60%+ reduction in non-production costs.
2. Not Right-Sizing Databases
Teams pick a database size during setup and never revisit it. We routinely find production databases running at <15% CPU utilization.
Fix: Monitor actual usage for 2 weeks, then right-size. Most databases can drop 1-2 instance sizes with zero performance impact.
3. Ignoring Data Transfer Costs
Cross-AZ and cross-region data transfer adds up fast. A single architectural mistake can cost £1,000+/month.
Fix: Audit your VPC topology. Keep high-traffic components (load balancers, databases, app servers) in the same AZ.
What This Actually Costs You
If you’re running a SaaS with £1M+ ARR and haven’t done a proper infrastructure audit in the last 6 months, you’re probably wasting £2,000-£5,000/month.
Over a year, that’s £24,000-£60,000 in pure waste.
How We Do This
Our Infrastructure Audit is a fixed-price, 2-week engagement:
- £3,000 fixed price
- 2 weeks delivery
- Complete cost optimization report
- We implement the top 5 quick wins
- ROI typically 10-15x in first year
If you’re spending £3,000+/month on AWS and haven’t audited your infrastructure recently, let’s talk.
Book a 30-Minute Discovery Call
We’ll review your AWS bill and tell you honestly if an audit makes sense for you.