Vocabulary for Cloud Cost Management

Cloud cost management — often called FinOps — has developed its own vocabulary over the past decade. Whether you are an engineer, a team lead, or a finance partner trying to understand a cloud invoice, you will encounter terms that are not self-explanatory and that carry precise meanings in business conversations.

This guide covers the core FinOps vocabulary in English, explains what each term means, and shows you how to use it naturally in meetings, reports, and Slack discussions.

---

Foundational Cost Concepts

Cloud spend / Cloud cost

These terms are interchangeable in most contexts. “Spend” is slightly more natural in business speech; “cost” is more common in technical writing.

• “Our cloud spend increased by 40% year over year.”
• “The team’s cloud cost for Q2 was £180,000.”
• “We need to understand what’s driving the spike in cloud spend before the quarterly review.”

Billing vs. cost allocation

Billing refers to the actual invoice from the cloud provider. Cost allocation refers to how that bill is distributed across teams, products, or cost centres within your organisation.

• “The AWS bill is consolidated at the organisation level. We use cost allocation tags to break it down by team.”

---

Savings and Optimisation Vocabulary

Rightsizing

Rightsizing means adjusting the size of a cloud resource — a virtual machine, a database instance, a container — to match its actual usage. An oversized instance costs more than necessary; an undersized one causes performance problems.

• “The cost analysis identified several instances running at under 10% CPU utilisation. Rightsizing those three instances alone would save £3,000 per month.”
• “We kicked off a rightsizing exercise after the annual cost review.”
• “Rightsizing is not a one-time task — usage patterns change, so we review it quarterly.”

Reserved instances (RIs)

Reserved instances are a pricing model where you commit to using a specific cloud resource for one or three years in exchange for a significant discount (typically 30–70% off on-demand pricing). Common in AWS, Azure, and GCP.

• “We purchased reserved instances for our baseline compute workloads — the ones that run 24/7 regardless of traffic.”
• “The team recommended converting on-demand instances to RIs for the production database tier.”
• “Reserved instances give you cost certainty but reduce flexibility — you’re locked in for the commitment period.”

Savings plans

Savings plans are a more flexible alternative to reserved instances. Instead of committing to a specific instance type, you commit to a dollar amount of compute spend per hour. The discount is applied automatically across any qualifying usage.

• “We switched from reserved instances to savings plans because they cover more resource types and don’t require us to predict exact instance configurations a year in advance.”
• “Compute savings plans apply to EC2, Lambda, and Fargate — which made them a better fit for our mixed workload.”
• “The FinOps team recommended a 1-year compute savings plan to cover our baseline spend.”

Spot instances

Spot instances (AWS) or preemptible VMs (GCP) are spare cloud capacity sold at steep discounts — often 60–90% off on-demand prices. The tradeoff is that the cloud provider can reclaim them with short notice.

• “We run batch processing jobs on spot instances — if a job gets interrupted, it restarts automatically.”
• “Spot instances are not appropriate for stateful, long-running workloads because of the interruption risk.”

---

Visibility and Accountability Vocabulary

Cost allocation tags

Tags are metadata labels applied to cloud resources (instances, buckets, databases). Cost allocation tags allow you to filter and aggregate your cloud bill by team, project, environment, or any dimension you define.

• “We require all engineers to tag every resource with team, environment, and project when they provision it.”
• “Cost allocation tags let us show each product team exactly how much of the cloud bill they own.”
• “The lack of consistent tagging has made it impossible to do accurate chargebacks — we’re enforcing the tagging policy starting next quarter.”

Chargeback

Chargeback is the practice of allocating cloud costs back to the business units, teams, or cost centres that incurred them — and actually charging them for it. This creates direct financial accountability.

• “We’ve implemented a chargeback model where each engineering team is billed for their actual cloud consumption.”
• “The finance team uses the chargeback data for internal P&L reporting.”
• “Chargeback requires accurate cost allocation tags. If the tags are missing, costs land in an unallocated bucket.”

Showback

Showback is the practice of showing teams what they are spending — providing visibility — without actually charging them for it. It is an intermediate step between full opacity and chargeback.

• “We started with showback before moving to full chargeback — it gave teams time to understand their costs and optimise before they were held financially accountable.”
• “The monthly showback report goes to all team leads. It shows cloud costs by team, trend over time, and the top cost drivers.”
• “Showback alone drove a 15% cost reduction in the first quarter — teams self-optimised once they could see the numbers.”

---

Waste and Inefficiency Vocabulary

Idle resources

Resources that are provisioned but not being used — or being used at negligible levels.

• “The cost audit found £8,000 per month in idle resources: 12 load balancers with no attached targets and 30 unattached EBS volumes.”
• “We implemented an idle resource cleanup script that runs weekly and sends a report to the Slack channel.”

Orphaned resources

Resources that were created for a purpose that no longer exists — a database for a decommissioned project, a VPC for a test environment that was never torn down.

• “We found orphaned resources from a proof-of-concept that was cancelled six months ago — the team forgot to tear down the environment.”

Overprovisioning

Allocating more capacity than a workload actually needs, either from over-caution or lack of monitoring.

• “The initial provisioning was conservative — we overprovisioned to be safe during launch. Six months later, the utilisation data shows we can scale down significantly.”

---

A Practical Vocabulary Reference

Term	One-line definition
Rightsizing	Matching resource size to actual usage
Reserved instances	Long-term commitment for discounted pricing
Savings plans	Flexible spend commitment for discounted pricing
Spot / preemptible	Cheap, interruptible spare capacity
Cost allocation tags	Metadata for distributing costs to teams
Chargeback	Billing teams for their actual cloud spend
Showback	Showing teams their spend without billing them
Idle resources	Provisioned but unused resources
Orphaned resources	Resources whose original purpose is gone
Overprovisioning	Allocating more capacity than needed

---

Using This Vocabulary in a Business Meeting

When presenting a cloud cost review to a mixed audience of engineers and business stakeholders, the vocabulary needs to bridge both worlds. Here is how a cloud cost discussion might sound:

“Our cloud spend for Q2 was £220,000, up from £160,000 in Q1. The increase is largely explained by the launch of the new product tier in June, which brought significant new compute and storage usage. We’ve identified three areas for optimisation: first, rightsizing — we have a significant number of instances running at under 20% CPU utilisation, and rightsizing those to the next instance size down would save approximately £12,000 per month. Second, we’re recommending moving our baseline compute to a one-year savings plan, which would reduce on-demand costs by around 30%. Third, we’re implementing mandatory cost allocation tags so we can move from showback to chargeback by Q4.”

That is approximately 130 words of coherent, professional FinOps English. The vocabulary is precise, the narrative is logical, and it works equally well in a written report or spoken in a meeting.