AWS Cost Explorer: The Complete Guide (And Its Blind Spots)

AWS Cost Explorer is the first tool every AWS user turns to when trying to understand their cloud bill. It's free (for the console UI — API access is $0.01/request), it's built in, and it contains a lot of genuine value. Knowing how to use it well separates teams that get actionable insights from teams that stare at a colorful chart and learn nothing.

This guide covers Cost Explorer comprehensively — every feature worth using, how to configure it for maximum insight, and then, critically, the five significant things Cost Explorer cannot do and what to use instead. Understanding the gaps matters as much as understanding the capabilities.

Getting Started: Enable Cost Explorer First

Cost Explorer must be enabled before it starts collecting data — and it only collects data from the date it's enabled, not retroactively. If you haven't enabled it, go to the AWS Billing console → Cost Explorer → Enable Cost Explorer. It takes up to 24 hours to populate initial data.

If you're running AWS Organizations, enable Cost Explorer at the management (payer) account level. This allows you to see spend across all member accounts from a single interface — critical for multi-account environments.

The Core Features Worth Mastering

Filtering and Grouping

Cost Explorer's power comes from its filter and group-by options. The most useful groupings:

• Group by Service: Your first view — shows the top-level breakdown of where your money goes (EC2, RDS, CloudFront, etc.)
• Group by Account: Essential for multi-account environments. See which accounts are driving costs.
• Group by Region: Identifies whether you're running resources in unexpected regions (often a sign of forgotten resources or misconfigured deployments).
• Group by Tag: The most powerful grouping once you have good tagging hygiene. Group by Environment (prod/staging/dev) to see your dev environment bill separately from production.
• Group by Instance Type: For EC2 specifically, shows exactly which instance types you're paying for — essential for rightsizing analysis.

Anomaly Detection

Cost Anomaly Detection (accessible from the Cost Explorer left nav) is an ML-powered feature that alerts you when spending patterns deviate unexpectedly. Unlike static budget alerts ($X threshold exceeded), anomaly detection adapts to your spending pattern — so it won't alert you for expected end-of-month spending spikes but will catch a new EC2 fleet someone provisioned without authorization.

Configure anomaly detection monitors for: (1) total AWS spend, (2) EC2 specifically (your largest service), and (3) any services where anomalies would be particularly costly (like RDS or Redshift). Set email or SNS alerts for anomalies above $50 to catch meaningful cost events without noise from small variations.

Rightsizing Recommendations

Cost Explorer → Recommendations → Rightsizing shows EC2 instances that are underutilized and recommends downsizing options with estimated monthly savings. The quality of these recommendations is good — they're based on 14 days of CloudWatch metrics for CPU utilization.

Limitations to know: the recommendations only look at CPU and network utilization, not memory (because CloudWatch doesn't collect memory metrics by default without the CloudWatch Agent). An instance can appear underutilized on CPU but be memory-constrained. Always verify memory usage before downsizing. Also, recommendations only cover EC2 — not RDS, ElastiCache, or other services.

Savings Plans and RI Recommendations

Cost Explorer provides purchase recommendations for both Compute Savings Plans and EC2 Instance Savings Plans. Navigate to Savings Plans → Recommendations and select your preferred commitment term (1 year vs 3 year) and payment option (no upfront vs partial vs all upfront).

The recommendations analyze your last 7 or 30 days of usage and calculate the optimal Savings Plan commitment that maximizes savings. One critical caveat: recommendations assume your current usage is representative of future usage. If you're planning to scale up or down significantly, adjust the recommendation accordingly.

For detailed guidance on when to buy RIs vs Savings Plans, see our full comparison guide.

The 5 Things AWS Cost Explorer Cannot Do

Cost Explorer is genuinely useful, but it has fundamental limitations that require you to use additional tooling once your FinOps practice matures.

Blind Spot 1: Cross-Account Aggregation at Scale

Cost Explorer in Organizations can show all accounts together, but its filtering and grouping capabilities don't scale elegantly to hundreds of accounts. You can't easily say "show me all EC2 spend tagged Environment=prod across my 50 production accounts, grouped by the Team tag." At 10+ accounts with complex tagging, native Cost Explorer becomes unwieldy for cross-account analysis. Dedicated FinOps tools like Hero Savings or CloudHealth are built for this use case.

Blind Spot 2: No 1-Click Remediation

Cost Explorer shows you what to fix but provides no mechanism to fix it. You see that a specific EC2 instance is a rightsizing candidate — great. Now you need to open the EC2 console, find the instance, plan the maintenance window, downsize it, monitor it. Cost Explorer's role ends at the recommendation. Every action requires manual follow-through. This is the primary reason optimization recommendations go unacted upon — the friction between seeing the opportunity and capturing it is too high.

Blind Spot 3: No Terraform / IaC Output

If you manage your infrastructure as code, Cost Explorer's recommendations are in "English" (e.g., "downsize i-0abc123 from m5.2xlarge to m5.xlarge") not in Terraform. There's no way to take a rightsizing recommendation and generate the corresponding Terraform change. Engineers have to manually translate every recommendation into code, test it, and apply it through their CI/CD pipeline. For IaC-managed environments, this gap significantly increases the effort required to act on recommendations.

Blind Spot 4: No Natural Language Queries

Cost Explorer's UI is powerful but requires knowing what you're looking for. You can't ask "why did my bill increase by $3,000 last week?" and get a direct answer — you have to manually navigate the filters, adjust the date range, group by different dimensions, and piece together the story yourself. This is a significant usability gap for non-technical stakeholders (finance teams, engineering managers) who need cost answers without becoming Cost Explorer power users.

Blind Spot 5: No Automation or Policy Enforcement

Cost Explorer is purely read-only visibility — it can't stop waste from happening, only help you see it after the fact. It won't alert you within hours when a new idle resource is created, can't automatically clean up orphaned EBS volumes, can't enforce tagging requirements on new resources, and can't automatically purchase Savings Plans when your baseline spend stabilizes. All of these automation capabilities require either custom engineering work (Lambda functions, AWS Config rules) or a dedicated cost management platform.

Building Your Cost Explorer Workflow

Here's the weekly Cost Explorer routine that covers 80% of the value with 20% of the effort:

1. Monday morning (5 min): Open Cost Explorer, view last 7 days vs previous 7 days. Look for any significant increases. If anomaly detection hasn't alerted you, a quick visual scan catches most surprises.
2. Monthly (30 min): Review rightsizing recommendations, Savings Plan recommendations, and month-over-month cost by service. Generate a cost summary for engineering leadership.
3. Quarterly (2 hours): Full RI/Savings Plan portfolio review. Are you hitting your coverage targets? Any recommendations to act on? Review tag coverage reports to identify untagged resources.