ViitorCloud
Home / Resources / Runbook

A Cloud FinOps 90-day runbook for AWS, Azure, and GCP.

2026 industry research puts cloud waste between 25 and 35% of total cloud spend. Most of the waste is recoverable. Holding the savings is the hard part. This is a runbook, not a discussion: phases run in sequence, activities inside each phase in parallel. It assumes one FinOps lead, two cloud engineers, a named finance partner, executive sponsorship, and a signed savings target.

Playbook / RunbookAuthor: Vishal Shukla, VP of TechnologyBFSI · Healthcare · Public SectorRelevant to: Cloud FinOps · Analytics
Key takeaways
Cloud waste is roughly 25–35% of total spend.
First-quarter realised savings are typically 20–35%; best-case 30–50%.
Three phases over thirteen weeks: quick wins → governance and tagging → operating model.
Day-90 success: realised savings vs the SOW, tagging compliance above 90%, and an operating cadence run by the client team that survives partner exit.
The runbook

Three phases, thirteen weeks.

Phase 1, Quick Wins

Weeks 1–4
Week 1

Estate audit and baseline

save 0%

Ingest AWS CUR, Azure Cost Management, and GCP Billing into one warehouse; 30 days of utilisation; top 20 cost drivers; baseline report plus a signed savings target in the SOW.

Week 2

Idle and orphaned resource cleanup

save 5–12%

Idle compute (under 5% CPU for 14 days), orphaned storage, zombie load balancers, idle databases, unused NAT gateways, unattached static IPs.

Week 3

Rightsizing on compute

save 15–25%

AWS Compute Optimizer, Azure Advisor, GCP Recommender; cross-check the business calendar; dev/test first; Kubernetes requests and limits via Kubecost.

Week 4

Commitment optimisation

30–60% on covered workloads

Audit RIs, Savings Plans, and CUDs (under 90% utilisation is a candidate); model the new posture against rightsized workloads; coverage target 70–80%.

Phase 2, Governance and Tagging

Weeks 5–8
Week 5

Tagging standard + enforcement architecture

foundation

Schema: Owner, CostCentre, Environment, Project, DataClassification. Enforcement layer: SCPs, Azure Policy, GCP Org Policy.

Week 6

Tagging rollout

target 90% by Week 8

Auto-tag what can be auto-tagged, tickets for the rest, policy-as-code for new resources.

Week 7

Cost attribution and showback

5–15% from visibility

Showback dashboard at team and workload level; monthly to leadership and finance.

Week 8

Commitment execution

30–60% lands by Week 12

Execute the Week 4 plan; schedule the first quarterly review.

Phase 3, Operating Model Rollout

Weeks 9–13
Week 9

Operating rhythm design

durability

Weekly FinOps + engineering review, monthly finance + engineering forecast (variance over 10% triggers deeper review), quarterly commitment and tag-schema review; name the owners.

Week 10

Anomaly detection and alerting

early warning

Native anomaly detection across the three clouds; service and team thresholds; an anomaly response runbook.

Week 11

Engineering enablement

5–10% over two quarters

Enablement sessions; cost gates in the pipeline (Infracost or native); embed the dashboards.

Week 12

AI workload hygiene

10–25% on AI workloads

GPU and inference utilisation (30–50% is commonly over-provisioned); rightsize GPU, batch to spot, vector store and embedding sizing; review monthly.

Week 13

Handover and retrospective

20–35% realised vs SOW

Handover to the Week 9 owners; final savings report; clean exit or a managed retainer.

Success at day 90

Realized savings vs the SOW (banked, not modelled) · tagging compliance above 90% · an operating cadence run by the client team with a named owner that survives partner exit. Two of three green is success; only one green means point-in-time savings without durable discipline.

PDF

Get the 7-page runbook

Includes the SOW template, savings calculator, tagging schema, and cadence template.

Want the runbook run for you, against a signed target?

The Cloud FinOps engagement writes the savings target into the SOW.

See the Cloud FinOps service