Cutting Cloud Bills While Scaling Revenue 5x

The Scaling Problem

Growth should improve efficiency. At ShopScale, it did the opposite.

• Revenue: ~$15M ARR
• Cloud costs: $20K → $80K/month
• Growth rate: 40% cost increase per quarter

The Hidden Risk

• Performance degrading despite higher spend
• Infrastructure complexity increasing
• Black Friday expected to break the system

This was not a scaling problem. It was an efficiency problem

Where the Money Was Going

We ran a deep cost audit across 90 days of AWS billing data.

Key Findings

Compute

• EC2 instances oversized
• ~40% unused capacity

Storage

• Old backups + unused snapshots
• No lifecycle policies
• Cost: ~$12K/month

Data Transfer

• Cross-region traffic with no business need
• Cost: ~$8K/month

Database Layer

• Missing indexes
• N+1 queries
• No read replicas

~$15K/month in wasted database compute

Optimisation Strategy

This was executed as a phased FinOps + engineering programme.

Phase 1: Cost Audit (Week 1)

• Analysed detailed billing data
• Identified cost drivers
• Built savings model
• Prioritised by ROI vs effort

Phase 2: Compute Optimisation (Weeks 2–3)

• Analysed CPU/memory usage (CloudWatch)
• Right-sized EC2 and RDS instances
• Introduced auto-scaling

Savings: $18K/month

Phase 3: Reserved Instances (Weeks 4–5)

• Identified stable workloads
• Purchased 1-year Reserved Instances
• Used spot instances for variable load

Savings: $12K/month

Phase 4: Storage Optimisation (Week 6)

• Backup compression (60% reduction)
• S3 Intelligent-Tiering enabled
• Removed 47 unused snapshots
• Log lifecycle policies implemented

Savings: $8K/month

Phase 5: Database Optimisation (Weeks 7–8)

• Added 12 high-impact indexes
• Eliminated N+1 query patterns
• Introduced read replicas

Savings: $6K/month

Phase 6: Data Transfer Optimisation (Week 9)

• Removed unnecessary cross-region transfers
• Introduced CDN (CloudFront)
• Reduced transfer volume by 70%

Savings: $4K/month

Phase 7: Application-Level Optimisation (Week 10)

• Redis caching layer
• Pagination for large datasets
• Image optimisation (WebP, responsive sizing)

Savings: $2K/month + performance gains

Cost Breakdown

Compute (EC2, RDS)
→ $32K → $18K ($14K/month saved)

Storage (S3, backups)
→ $18K → $6K ($12K/month saved)

Data transfer
→ $18K → $4K ($14K/month saved)

Database overhead
→ $15K → $4K ($11K/month saved)

Total
→ $83K → $32K ($51K/month saved)

Total Savings

$51K/month$612K/year~60% reduction

Performance Improvements

Cost optimisation did not reduce performance. It improved it.

• Page load time: 4.2s → 2.3s (45% faster)
• Query latency: 250ms → 120ms
• Concurrent users: 5,000 → 50,000

Black Friday Outcome

• Handled 10× traffic spike
• No infrastructure scaling required
• Zero outages

System was finally right-sized, not overbuilt

Business Impact

The savings unlocked strategic capacity:

• +4 engineering hires funded
• +6% improvement in profit margin
• Increased investment in product development
• Stronger pricing competitiveness

What Made the Difference

1. Baseline visibility firstDecisions driven by real usage data
2. Right-size before committing spendAvoid locking inefficiency into Reserved Instances
3. Application-level fixes matter mostQuery optimisation delivered disproportionate gains
4. Automated cost controlsPrevents regression over time
5. Continuous monitoringCost efficiency becomes an ongoing discipline

The Key Insight

Cloud cost problems are rarely caused by infrastructure alone. They are a combination of:

• architecture decisions
• inefficient queries
• lack of cost governance

FinOps + engineering must work together

Final Outcome

ShopScale transitioned from:

• Runaway cloud spend → controlled cost model
• Degrading performance → optimised experience
• Scaling risk → scalable infrastructure

Result:
A platform that is:

• faster
• cheaper
• capable of handling growth

Optimising Cloud Spend at Scale?

Intagleo helps companies reduce infrastructure costs while improving performance through engineering-led optimisation, not just cost cutting.

Start a conversation