Cost Analytics

Starter+Enterprise

Cost Analytics connects to Cloud Cost Explorer to give you a complete picture of your cloud spend — broken down by service, region, account, and resource tag — alongside actionable recommendations for reducing waste. Instead of switching between the cloud Console, spreadsheets, and third-party FinOps tools, everything you need to understand and optimise your cloud bill is available directly in Calabi.

Key Capabilities

Capability	Description
Spend breakdown	Filter and group costs by service, region, account, linked account, or tag
Month-over-month trends	Line and bar charts comparing current vs. prior periods
Rightsizing recommendations	Identify underutilised EC2, idle RDS, and oversized EBS
Cost anomaly detection	Alerts when daily spend exceeds expected patterns
Savings Plans coverage	See what fraction of your compute is covered by Savings Plans
Reserved Instance coverage	Identify on-demand usage that could be replaced by RIs
Tag coverage report	Surface untagged resources driving unattributed spend

Spend Breakdown

The Spend Breakdown view is the primary cost exploration interface. It renders a stacked bar chart for the selected time range and a detailed table beneath it.

Grouping dimensions

Dimension	Example use case
Service	How much is EC2 vs. S3 vs. RDS this month?
Region	Which regions are driving the most spend?
Linked account	How does spend compare across AWS Organization member accounts?
Usage type	Break EC2 cost into compute, data transfer, and EBS components
Resource tag	Attribute costs to teams, projects, or environments via tags

Time range presets

Preset	Description
This month	Month-to-date (MTD)
Last month	Full previous calendar month
Last 3 months	Rolling 90-day view
Last 6 months	Rolling 180-day view
Last 12 months	Full year view
Custom	Arbitrary start and end date

Data freshness

Cloud Cost Explorer data has a 24-hour lag. Charges incurred today will appear in Calabi the following day.

Month-over-Month Trends

The Trends tab shows your total cloud spend as a line chart across the last 12 months, with the current month shown as a partial bar. Key trend indicators:

MoM change (%) — percentage change vs. the prior month
3-month moving average — smoothed trend line to filter month-end noise
Projected month-end — linear projection of the current month's final cost based on daily spend rate

You can overlay a second dimension (e.g., EC2 + RDS) to compare two cost lines on the same chart.

Rightsizing Recommendations

Calabi analyses CloudWatch utilisation metrics alongside Cost Explorer data to surface rightsizing opportunities. Recommendations are grouped by resource type.

EC2 rightsizing

Finding type	Trigger condition	Potential saving
Underutilised instance	Average CPU < 5% and network < 5 MB/day for 14 days	Switch to a smaller instance type
Idle instance	Average CPU < 1% for 14 days	Stop or terminate the instance
Oversized memory	Memory utilisation < 10% for 14 days (requires CloudWatch agent)	Move to a memory-optimised tier

Each recommendation includes:

Current instance type and monthly cost
Recommended instance type and monthly cost
Estimated monthly saving
Last 14 days of CPU and network utilisation charts
A direct link to resize the instance in the cloud Console

RDS rightsizing

Finding type	Trigger condition
Idle cluster	Zero connections for 7 days
Oversized instance	Average CPU < 5%, FreeableMemory > 80% for 14 days
Single-AZ in production	Non-dev instance with `multi_az = false`

EBS optimisation

Finding type	Trigger condition
Unattached volume	`state = available` for more than 14 days
gp2 to gp3 upgrade	Any gp2 volume (gp3 is cheaper and faster by default)
Oversized provisioned IOPS	Provisioned IOPS utilisation < 20% for 14 days

Cost Anomaly Detection

Calabi applies a statistical model to your daily spend time series and alerts you when actual spend deviates significantly from the expected range.

How it works

Calabi trains on 90 days of historical daily spend per service and account.
For each new day, it computes an expected spend range (lower and upper bound).
If the actual daily spend exceeds the upper bound by more than the configured threshold, an anomaly is raised.
Anomaly alerts are delivered to your configured notification channels (Slack, PagerDuty, email).

Anomaly alert example

[ANOMALY] EC2 spend spike detected
Account:    123456789012 (production)
Service:    Amazon EC2
Date:       2026-04-07
Expected:   $1,200 – $1,450
Actual:     $3,820
Deviation:  +163%

Top driver: us-east-1 — 47 new p3.2xlarge instances launched at 14:23 UTC

Savings Plans and Reserved Instance Coverage

Savings Plans coverage

The coverage chart shows what fraction of your EC2, Fargate, and Lambda compute is covered by a Savings Plan commitment vs. on-demand pricing. The uncovered (on-demand) portion is highlighted in orange — these are the hours where a Savings Plan would reduce costs.

Metric	Description
Coverage %	Percentage of eligible compute hours covered by a Savings Plan
On-demand cost (uncovered)	Monthly cost of the uncovered compute hours
Recommended commitment	Daily Savings Plans commitment that would achieve 80% coverage
Estimated annual saving	Projected savings at the recommended commitment level

Reserved Instance coverage

Metric	Description
RI coverage %	Percentage of RDS or ElastiCache instance hours covered by RIs
On-demand hours	Hours running at on-demand price that could be RI-covered
Recommended RI purchase	Instance type and term recommendation

Tag Coverage Report

Untagged resources make it impossible to attribute costs to teams or projects. The Tag Coverage report shows:

Which resources have missing or incomplete tags
What fraction of total spend is unattributable (untagged)
The estimated unattributable cost for the current month

You can define which tag keys are required (e.g., env, team, project) in Configure under Tag Policy.

Example SQL Queries

These queries can be run directly in Query Resources.

Total daily spend by service (last 30 days)

SELECT
  service,
  usage_start_date,
  SUM(unblended_cost_amount) AS daily_cost_usd
FROM
  aws_cost_by_service_daily
WHERE
  usage_start_date >= CURRENT_DATE - INTERVAL '30 days'
GROUP BY
  service, usage_start_date
ORDER BY
  usage_start_date DESC, daily_cost_usd DESC;

Top 10 most expensive services this month

SELECT
  service,
  ROUND(SUM(unblended_cost_amount)::NUMERIC, 2) AS month_to_date_cost_usd
FROM
  aws_cost_by_service_daily
WHERE
  usage_start_date >= DATE_TRUNC('month', CURRENT_DATE)
GROUP BY
  service
ORDER BY
  month_to_date_cost_usd DESC
LIMIT 10;

Month-over-month EC2 spend comparison

SELECT
  DATE_TRUNC('month', usage_start_date)          AS month,
  ROUND(SUM(unblended_cost_amount)::NUMERIC, 2)  AS ec2_cost_usd
FROM
  aws_cost_by_service_daily
WHERE
  service = 'Amazon EC2'
  AND usage_start_date >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY
  DATE_TRUNC('month', usage_start_date)
ORDER BY
  month;

Spend by tag (team)

SELECT
  tags ->> 'team'                                AS team,
  ROUND(SUM(unblended_cost_amount)::NUMERIC, 2)  AS monthly_cost_usd
FROM
  aws_cost_by_service_daily
WHERE
  usage_start_date >= DATE_TRUNC('month', CURRENT_DATE)
  AND tags ->> 'team' IS NOT NULL
GROUP BY
  tags ->> 'team'
ORDER BY
  monthly_cost_usd DESC;

Age Reports — Find stale resources contributing to unnecessary storage costs
Security Posture — Identify misconfigured resources that may also be driving unexpected spend
Configure — Set up tag policies and anomaly alert channels

Key Capabilities​

Spend Breakdown​

Grouping dimensions​

Time range presets​

Month-over-Month Trends​

Rightsizing Recommendations​

EC2 rightsizing​

RDS rightsizing​

EBS optimisation​

Cost Anomaly Detection​

How it works​

Anomaly alert example​

Savings Plans and Reserved Instance Coverage​

Savings Plans coverage​

Reserved Instance coverage​

Tag Coverage Report​

Example SQL Queries​

Total daily spend by service (last 30 days)​

Top 10 most expensive services this month​

Month-over-month EC2 spend comparison​

Spend by tag (team)​

Related Pages​