← All articles

FinOps 101: How a Small Team Can Implement Cost Discipline This Quarter

What FinOps Means for a Small Team

FinOps is the discipline of treating cloud spend like any other operational cost: measure it, allocate it, and continuously improve. For a team of 3‑10 engineers the goal isn’t a full‑blown finance department; it’s a lightweight loop that surfaces waste fast enough to act on before the next billing cycle. The core loop is: 1. Collect – pull cost data daily. 2. Analyze – map spend to owners and services. 3. Act – remediate idle resources, adjust sizing, or apply discounts. 4. Learn – capture the decision and repeat. The following sections give a concrete 90‑day plan that can be executed with the tools already available in AWS, GCP, and Azure.

1. Establish a Cost Baseline with Tagging and Allocation Rules

a. Enforce a Minimal Tag Set

All resources should carry at least three tags: - owner – the engineer or team responsible. - environment – prod, staging, or dev. - project – a short identifier for the product or feature. Create a tag policy to prevent drift. Example for AWS:

aws organizations create-policy \
  --content file://tag-policy.json \
  --description "Require owner, environment, project tags" \
  --name "EnforceFinOpsTags" \
  --type SERVICE_CONTROL_POLICY

For GCP, use an Organization Policy:

gcloud org-policies enable-enforce \
  --project=my-project \
  --constraint=constraints/compute.requireOsLogin

(Replace with the appropriate tag constraint.) Azure uses policy definitions:

az policy definition create \
  --name "require-tags" \
  --rules tag-policy.json \
  --mode All

b. Pull the First Cost Report

Run a one‑time cost export to set the baseline. - AWS:

aws ce get-cost-and-usage \
  --time-period Start=$(date -d '-30 days' +%Y-%m-%d),End=$(date +%Y-%m-%d) \
  --granularity DAILY \
  --metrics UnblendedCost \
  --group-by Type=DIMENSION,Key=TAG

• GCP:

gcloud beta billing accounts list
ACCOUNT=$(gcloud beta billing accounts list --format='value(name)' --filter='open:true')
gcloud beta billing budgets list --billing-account=$ACCOUNT

• Azure:

az consumption usage list \
  --start-date $(date -d '-30 days' +%Y-%m-%d) \
  --end-date $(date +%Y-%m-%d) \
  --query "[?tags.owner].{date:date, cost:pretaxCost, owner:tags.owner}" \
  -o table

Store the CSV/JSON in a shared bucket (e.g., s3://finops-baseline/ or gs://finops-baseline/). This file becomes the reference point for all future variance analysis.

2. Set Up a Weekly Cost Review Cadence

a. Create a Simple Dashboard

Use a free tool like Grafana or Google Data Studio to visualize the tag‑grouped spend. Connect the data source to the exported CSV files. - In Grafana, add a CSV datasource plugin, point it at the S3 bucket, and build a panel that groups by owner.

b. Define a Review Meeting Agenda

import boto3, json, os

def handler(event, context):
    ce = boto3.client('ce')
    resp = ce.get_cost_and_usage(
        TimePeriod={'Start': os.getenv('START'), 'End': os.getenv('END')},
        Granularity='DAILY',
        Metrics=['UnblendedCost'],
        GroupBy=[{'Type':'DIMENSION','Key':'TAG'}]
    )
    # write to S3
    s3 = boto3.client('s3')
    s3.put_object(Bucket='finops-reports', Key='weekly.json', Body=json.dumps(resp))
    # post to Slack
    # ... omitted for brevity ...

Replace the environment variables with the appropriate dates each run.

3. Automate Alerts and Right‑Sizing Recommendations

a. Set Threshold Alerts

• AWS Budgets:

aws budgets create-budget \
  --account-id 123456789012 \
  --budget file://budget.json

budget.json should contain a BudgetLimit and an AlertThreshold of 80 %. - GCP Billing Alerts:

gcloud beta billing budgets create \
  --billing-account=$ACCOUNT \
  --display-name="FinOps Alert" \
  --budget-amount=5000 \
  --threshold-rule=threshold=0.8,spend_basis=CURRENT_SPEND

• Azure Cost Alerts:

az consumption budget create \
  --budget-name "FinOpsAlert" \
  --amount 5000 \
  --time-grain Monthly \
  --category Cost \
  --notifications "operator=Email;threshold=80"

When an alert fires, the responsible owner tag is extracted and a ticket is auto‑created (e.g., via GitHub Issues API).

b. Use Built‑In Right‑Sizing Tools

# AWS EC2
aws ec2 stop-instances --instance-ids i-0123456789abcdef0
# GCP Compute
gcloud compute instances stop my-instance --zone=us-central1-a
# Azure VM
az vm deallocate --resource-group rg-prod --name vm-prod-01

• Convert on‑demand to Savings Plans / Committed Use when a workload shows >30 % stable usage over 3 months. Use the cost explorer to identify the SKU and purchase via the console or CLI.

4. Institutionalize FinOps with Simple Governance

1. Document the Tag Policy in the repo’s README.md and protect it with CODEOWNERS.
2. Add a “FinOps Checklist” to every pull request that creates cloud resources. Example checklist item: - [ ] Resource hasowner,environment, andprojecttags - [ ] Estimated monthly cost < $100 (or approved by manager)
3. Rotate the Review Lead each sprint so knowledge spreads across the team.
4. Reward Cost‑Saving Actions: a small quarterly bonus or public shout‑out for the engineer whose ticket saved the most dollars.

5. Quick Wins You Can Implement This Quarter

• Delete unattached Elastic IPs, static IPs, and NAT Gateways using the CLI filters shown in the earlier sections.
• Enable S3 Intelligent‑Tiering for infrequently accessed objects:

aws s3api put-bucket-lifecycle-configuration \
  --bucket my-bucket \
  --lifecycle-configuration file://intelligent-tiering.json

• Turn on GCP’s “Idle Disk” recommendations to shrink persistent disks that are >80 % empty.
• Consolidate Azure Dev/Test subscriptions under the Azure Dev/Test pricing tier to get up to 55 % discount automatically. These actions typically show results in the next billing cycle without any architectural changes.

FinOps platforms like CloudBudgetMaster automate the entire loop: they continuously ingest cost data from AWS, GCP, and Azure, map spend to your custom tags, surface idle resources with an estimated dollar impact, and push alerts to Slack or ticketing systems. The result is a real‑time, organization‑wide view of waste without manual scripting.