AWS Trusted Advisor vs Compute Optimizer vs Cost Explorer: What Each One Actually Finds

AWS gives you three native tools that all claim to help with cost, and the "my provider says I'm already optimized" confidence usually comes from having glanced at one of them. The problem is that Trusted Advisor, Compute Optimizer, and Cost Explorer do genuinely different jobs, and no single one gives you the full picture. Most comparisons just line up their feature checkboxes. Here is what each actually finds, where each is blind, and how they fit together.

Cost Explorer: what you spent and where

Cost Explorer is the visualization and reporting layer. It answers "where is the money going": spend by service, by tag, by account, over time, with forecasting.

• Finds: trends, spikes, which service or team is growing, and it surfaces Reserved Instance and Savings Plans recommendations based on your usage history.
• Blind to: whether individual resources are right-sized or idle. Cost Explorer tells you EC2 is your biggest line; it does not tell you that a specific instance is at 4% CPU. It is the map, not the inspection.

Use it for the big-picture question and for commitment purchase recommendations. Do not expect it to point at individual waste.

Compute Optimizer: is each resource the right size

Compute Optimizer is the machine-learning rightsizing engine. It analyzes utilization metrics and recommends a better configuration per resource.

• Finds: over- and under-provisioned EC2 instances, Auto Scaling groups, EBS volumes, and Lambda functions, with a concrete recommended target and the projected savings. It is the tool that says "this m5.xlarge should be an m5.large."
• Blind to: things that are not a sizing question. It will not tell you a resource is entirely unused (that is idle/orphan territory), and its scope is limited to the resource types it supports. It also needs enough utilization history to be confident.

Use it for the rightsizing question specifically. Turn it on (it is free), because it produces the most actionable per-resource dollar figures of the three.

Trusted Advisor: broad checks across pillars

Trusted Advisor runs a catalogue of best-practice checks across cost, security, performance, fault tolerance, and service limits. On cost, it flags things like idle load balancers, underutilized EBS volumes, and idle RDS instances.

• Finds: a broad sweep of best-practice violations, including some idle-resource cost checks the other two do not cover.
• Blind to / limited by: the free tier only exposes a subset of checks; the full cost checks require a Business or Enterprise Support plan. And its checks are threshold-based best practices, not deep per-resource rightsizing analysis.

Use it as the broad hygiene sweep, but know that on the free tier you are seeing a fraction of what it can flag.

The honest gap: none of them covers everything

Here is what the "we're already optimized" crowd misses. Each tool has a scope, and the union of all three still leaves gaps:

• Cost Explorer: spend visibility + commitment recommendations.
• Compute Optimizer: rightsizing for a specific set of resource types.
• Trusted Advisor: broad best-practice checks (subset free).

None of them, on its own, gives you a complete "here is every dollar of recoverable waste across every resource type and every cloud" answer. They do not cover orphaned resources thoroughly, they are AWS-only, and the coverage is split across three consoles with different scopes and access tiers. That is exactly the gap third-party FinOps tools exist to fill: one continuous view across resource types and clouds with the dollar figure attached (which is the category ZopNight sits in). The native tools are a real and free starting point; they are not the whole map.

How to actually use the three together

1. Cost Explorer to see the big rocks and grab RI/SP recommendations.
2. Compute Optimizer (turn it on, it is free) for per-resource rightsizing with concrete targets.
3. Trusted Advisor for the broad hygiene sweep, understanding the free tier is partial.
4. Then look for what all three miss: orphaned and idle resources, cross-account and cross-cloud waste, and anything outside Compute Optimizer's supported types.

The take

"My provider says I'm optimized" usually means "I looked at one of three tools that each see a different slice." Cost Explorer shows spend, Compute Optimizer rightsizes specific resources, Trusted Advisor runs broad checks (mostly behind a support plan). Use all three, then be honest that even together they leave gaps. Optimized is not a checkbox one console gives you; it is the union of these plus the waste none of them catches.

Which of the three do you actually check, and has it ever given you a false sense of "we're fine"? For me it was Cost Explorer, a clean-looking trend line hid a pile of idle resources that only Compute Optimizer and a manual sweep surfaced.