Do we need a FinOps platform like CloudHealth or Cloudability?

Not below ~£20k/month spend. Native tools plus a dashboard suffice. Above that, the platforms pay for themselves quickly.

How much can a typical SME save?

15–30% in the first two quarters of disciplined FinOps. Sustained 5–10% annual savings after that through ongoing optimisation.

Should engineers see costs?

Yes. Showback or chargeback to engineering teams is the single biggest cultural lever. People optimise what they can see.

Are reserved instances worth it?

For steady-state production, almost always. For dev/test or variable workloads, often not – spot/preemptible or savings plans usually win.

How does FinOps fit with cloud migration?

Build FinOps into the migration project, not after. Retro-fitting tags and ownership is painful. Day-one tagging policy avoids 18 months of clean-up.

FinOps for SMEs: How to Stop Cloud Costs Drifting After Migration

Why cloud costs drift

Engineers create test resources and forget them
Production capacity scales up to meet a spike and never scales back
Snapshots and old backups accumulate
Premium services chosen by default when standard tiers would suffice
Data egress unmodelled until the invoice arrives

The minimum viable FinOps

1. Cost ownership

Every workload, every environment, every subscription has a named owner who reads the monthly cost report. If nobody owns it, nobody optimises it.

2. Tagging policy

cost-centre – aligned to your finance chart of accounts
owner – named person, not a team mailbox
environment – prod / staging / dev / sandbox
application – the product or service it supports
data-classification – ties to security and DLP rules

Enforce tags at creation via Azure Policy, AWS Service Control Policies or GCP Org Policy. Block untagged resources from being created.

3. Budgets and alerts

Monthly budget per cost-centre, with 50/75/90/100% email alerts
Hard-stop or auto-throttle on non-production at 100%
Quarterly budget review aligned with sales forecast

4. Commitments and reservations

1- or 3-year Reserved Instances / Savings Plans for steady-state production compute
Storage tier rightsizing (hot/cool/archive) by access pattern
Annual licence consolidation review (Microsoft EA, AWS Enterprise Discount)

5. Zombie hunting

Monthly sweep of: unattached disks, unused public IPs, idle load balancers, snapshots older than the retention policy, dev environments with no recent activity, orphaned data factory pipelines.

The optimisation checklist

Right-size VMs against actual CPU and memory utilisation
Move dev/test off premium SKUs
Use spot/preemptible instances for fault-tolerant workloads
Lifecycle policies on object storage (S3, Blob, GCS)
Auto-shutdown of dev environments outside working hours
Review egress paths – cross-region and internet egress are the silent killers
Database reserved capacity and serverless tiers where they apply

Reporting cadence

Daily: automated anomaly detection
Weekly: 15-minute FinOps stand-up with owners
Monthly: cost-centre report + optimisation review
Quarterly: commitment refresh + leadership summary

Tooling

Most SMEs do not need a commercial FinOps platform until the cloud spend passes ~£20k/month. Azure Cost Management, AWS Cost Explorer, GCP Billing reports and a single Looker / Power BI dashboard are usually enough.

How FinOps connects to security

Tagging is also security gold. A cloud services inventory for Cyber Essentials falls out of a good tagging policy. Zombie resources are often unpatched and forgotten – removing them improves both bill and risk.

Common SME mistakes

Tagging policy without enforcement (everyone ignores it)
Budgets with no alerts (no signal until invoice arrives)
Optimisation as a one-off project (drift returns within 6 months)
Commitments before usage stabilises (lock-in to over-capacity)
No engineering accountability for cost (finance owns it alone)