AWS Cost and Cloud Health Assessment
There is no separate free cost assessment to book. Cost and cloud-health questions are already in scope for the existing Free AWS Assessment — bring your Cost Explorer view on screen and a senior AWS engineer will walk it with you, live, for one hour. What follows on this page is the methodology behind the paid, written depth: which AWS data sources we read, what each one actually tells you and where it runs out, how we separate a measured fact from a rule-of-thumb from a practitioner’s judgment call, and the prioritized backlog that comes out the other end. The paid depth is a scoped Consulting engagement — priced individually after scoping, not a published rate.
Free path: the existing one-hour Free AWS Assessment, verbal only. The paid depth below is a scoped Consulting engagement — no standard public price.
The five dimensions examined
- Cost visibility and governance — is spend tagged and attributable, is there a usage-report export configured, do budgets and alerting exist.
- Rate and commitment efficiency — on-demand versus committed usage, and whether available commitment recommendations have been acted on.
- Resource right-sizing and waste — over-provisioned or idle compute, storage, and database resources identified by measurement, not guesswork.
- Reliability and resilience (non-security) — Multi-AZ posture, backup coverage, and service-limit headroom, where the underlying check is available (plan-gated, not universal).
- Operational and architectural health — tagging discipline, account and region structure, and Well-Architected Framework review coverage where a client has run one. Uptempo does not conduct security-control scanning here — that is the separate AWS Security Posture Assessment.
Each dimension resolves to one of four plain bands — Strong, Adequate, Needs attention, or Insufficient evidence (not scored) — never a single composite number.
Evidence sources, what each yields, and its gaps
Scroll sideways to compare
Source | What it actually yields | Where it runs out |
|---|---|---|
AWS Cost Explorer | Cost and usage by service, account, tag, and region; month-over-month trend; rightsizing, Savings Plans, and Reserved Instance recommendations. | Cross-account and commitment recommendations are partial on a member-account-only engagement — marked partial, never inferred. |
Cost and Usage Report (CUR) / AWS Data Exports | The authoritative line-item billing record — untagged-spend percentage, per-team allocation, amortized commitment coverage. | Most organizations have no export configured going in, and a new export has no history — marked "requires the CUR export — not available yet," not silently skipped. |
AWS Compute Optimizer | Measured, ML-based rightsizing findings for EC2, EBS, Lambda, ECS on Fargate, and RDS, with a projected-savings range attached to each finding. | Needs to be enabled first; resources below AWS's data-volume threshold return "insufficient data" rather than a finding — eligible vs. ineligible counts are shown, never a false zero. |
AWS Trusted Advisor | Best-practice checks across cost, fault tolerance, service limits, and operational excellence. | Full checks require Business Support or higher. On Basic or Developer support, the reliability dimension is marked "partial coverage — full checks require a higher support plan." |
AWS Budgets | Whether cost governance exists at all — budgets, thresholds, alerting. | Its absence is itself a finding, not a data gap. |
AWS Well-Architected Tool | A structured, facilitated review against AWS's own framework, producing risk items and an improvement plan. | Human-facilitated, not an automated read — only appears where a client has already run one, and its output is always flagged as practitioner judgment (see Validation), never folded into an automated band. |
Any evidence collection is scoped, disclosed, and read-only, using access the client controls and can revoke — the same access-and-continuity model described on Security and Trust. Uptempo does not claim, here or anywhere, to run or host a cost-scanning product; the sources above are AWS's own, read directly.
How this actually runs
- Start with the free hour.
Book the Free AWS Assessment and bring your Cost Explorer view, a recent bill, or the line item that doesn't make sense. Findings are verbal, on the call — nothing written, no follow-up summary.
- If the free hour points to a written, quantified read,
the next step is scoping a Consulting engagement. Scope, access needs, and price are agreed in writing before any evidence collection starts.
- Evidence collection is scoped to what's actually available.
If a CUR export or Compute Optimizer enrollment doesn't exist yet, setting it up is one of the first scoped steps.
- The coverage matrix comes first, not last.
Before any finding is presented, the deliverable states exactly what was collected, what wasn't, and why.
- Findings become a prioritized backlog,
not a single grade. Each item links to the measured finding that justifies it.
- Handoff and routing.
A scoped Consulting build against the backlog, ongoing Support and Advisory capacity, a client-owned backlog item, or no further engagement — all legitimate outcomes.
What stays yours
Ownership stays with the client. Any read access is scoped, disclosed, and revocable by the client at any time — see Security and Trust for the general access-and-continuity model this follows. Evidence collected for one engagement is not retained in a shared database or used to build a cross-client benchmark corpus.
Explicitly excluded: a guaranteed or estimated aggregate savings figure, or a percentage claim — any dollar range shown is tied to one specific, cited AWS-projected finding, never a portfolio-wide total. No branded score, index, or single numeric grade of any kind. No claim of a comprehensive or complete read without stating the coverage boundaries that produced it. No security-control scanning (that's the separate AWS Security Posture Assessment). No managed operations, monitoring, or ongoing bill oversight — that capacity, if wanted, is Support and Advisory, a separate engagement. No audit, certification, or compliance-assurance claim of any kind.
How a finding earns its place
Every line in the deliverable carries one of three tags. Measured — a direct fact read from an AWS API or report (for example, "Compute Optimizer classifies N EC2 instances as over-provisioned"). Heuristic — a documented, stated rule applied to measured data; the rule and its threshold are always shown, never left implicit. Judgment — a practitioner's interpretation that needs the client's own confirmation (for example, a Well-Architected risk item); judgment-class findings are always labeled as such and never folded into an automated band.
The coverage matrix is mandatory and sits up front, not buried in an appendix: a table of dimension by evidence source, marked collected, unavailable (with the stated reason), not applicable, or access-denied. A dimension with insufficient evidence is marked "Insufficient evidence — not scored," never defaulted to a passing or failing grade. No line anywhere in the deliverable uses the words "comprehensive," "complete," or "full audit" — the coverage matrix is what replaces that kind of claim with a stated, honest boundary.
Where this connects
AWS Migration Strategy Workshop · AWS Security Posture Assessment · Assessments hub · Consulting · Support and Advisory
Start with the free hour
Cost and cloud-health questions are in scope for the Free AWS Assessment today. If a written, quantified read is the right next step, we scope it from there — as a Consulting engagement, priced after scoping.
Book Your Free AWS Assessment Talk about a scoped engagement