Purpose
Produce a defensible, weighted comparison of AWS, Azure, and GCP against a specific workload portfolio and a ranked primary/secondary recommendation with rationale — replacing vendor-marketing-driven decisions and gut feel with a scoring matrix an architecture review board can actually approve.
When to Use
Invoke this skill when the user asks to:
- Decide between AWS, Azure, and GCP for a workload or portfolio
- Justify a cloud platform selection to a steering committee or board
- Compare cloud providers against compliance, geography, or data-residency requirements
- Evaluate provider fit for an AI/ML, data, or HPC-heavy workload
- Decide whether to standardize on one provider or pursue multi-cloud
- Build the provider-selection section of a cloud transformation strategy
Inputs Required
Before starting, collect:
- Workload profile: Primary workload types — web/app compute, batch/HPC, data warehousing, ML/AI training and inference, transactional databases, etc.
- Scale and growth: Approximate current footprint (VMs, storage TB, monthly spend if known) and expected 24-month trajectory
- Geography: Regions where workloads must run and where data must reside
- Compliance requirements: Frameworks in scope — HIPAA, PCI-DSS, FedRAMP, GDPR, SOC 2, data sovereignty mandates
- Existing vendor relationships: Current Microsoft EA, Google Workspace, Oracle, SAP, or AWS commitments; existing licensing leverage
- Internal skills: Where the team's current expertise sits (e.g., .NET/Windows, open-source/Kubernetes, data engineering)
- Strategic priorities: The 1–3 dimensions that matter most (cost, AI/ML depth, compliance, time-to-value, exit flexibility)
If strategic priorities are not given, default to equal weighting across the 8 dimensions and state the assumption explicitly.
Instructions
Step 1: Establish the Weighted Scoring Dimensions
Score each provider 1–5 on these 8 dimensions. Assign each a weight (summing to 100%) based on the user's stated strategic priorities; if none given, weight all dimensions at 12.5%.
- Compute & core infrastructure breadth — instance variety, autoscaling, container/serverless maturity
- Storage & data services — object/block storage, managed databases, data warehouse maturity
- ML/AI capability — managed training, foundation models, MLOps tooling, accelerator availability
- Geographic coverage & data residency — regions/availability zones in required geographies, sovereign cloud options
- Compliance & certifications — coverage of the frameworks in scope, attestation availability
- Cost & commercial fit — pricing model alignment, existing commitments/discounts, licensing leverage
- Ecosystem & vendor alignment — fit with existing software stack and enterprise agreements
- Skills & operational readiness — alignment with internal team expertise and hiring market
State the final weights in a one-line legend before scoring.
Step 2: Score Each Provider Against the Matrix
For AWS, Azure, and GCP, assign a 1–5 score per dimension with a one-sentence justification tied to this workload portfolio — not generic provider descriptions. Calibrate against known provider strengths:
- AWS: broadest service catalog, deepest region coverage, strongest general-purpose maturity
- Azure: strongest fit for Microsoft-centric estates, enterprise agreements, hybrid (Azure Arc), and regulated/sovereign workloads
- GCP: strongest data/analytics (BigQuery) and ML/AI (Vertex, TPUs) depth, Kubernetes-native posture
Avoid scoring purely on marketing claims — tie each score to a concrete capability relevant to the inputs.
Step 3: Calculate Weighted Totals
Present a scoring table: rows = 8 dimensions, columns = weight, AWS raw, Azure raw, GCP raw. Compute weighted totals (raw × weight) per provider and a final weighted score out of 5.00 for each. Rank the three.
Step 4: Identify Tipping-Point Factors
Beyond the math, surface 2–4 factors that could override the numeric ranking:
- Hard constraints: A required compliance certification or region a provider lacks is disqualifying regardless of score
- Lock-in and exit risk: Note where the leading choice creates meaningful switching cost
- Commercial leverage: Existing EAs or committed-spend discounts that materially shift effective cost
- Skills gap timeline: Whether the recommended provider requires reskilling that delays time-to-value
Step 5: Make the Recommendation
- Primary provider: Named choice with a 3–4 sentence rationale referencing the top weighted dimensions and any tipping-point factors
- Secondary provider: Named fallback with the specific scenario in which it would win (e.g., "if FedRAMP High becomes a near-term requirement")
- Multi-cloud guidance: A direct opinion on whether to standardize on one provider or split — only recommend multi-cloud when a workload-specific advantage (e.g., GCP for analytics) justifies the added operational cost. Default against multi-cloud unless the inputs warrant it.
Output Format
- Structure: Weights legend → scoring matrix table → tipping-point factors → ranked recommendation
- Length: 900–1,300 words
- Format: Prose for rationale; a single scoring table for the matrix; bulleted tipping-point factors
- Sections: Scoring Methodology & Weights → Provider Scoring Matrix → Tipping-Point Factors → Recommendation (Primary / Secondary / Multi-Cloud Guidance)
- Tone: Enterprise architect advising a decision board — decisive, evidence-led, and willing to name disqualifying constraints
Quality Checklist
Before delivering output, verify:
- [ ] All 8 dimensions are scored for all 3 providers (24 cells minimum)
- [ ] Dimension weights are explicitly stated and sum to 100%
- [ ] Weighted totals are calculated and the three providers are ranked numerically
- [ ] Each score has a justification tied to the user's actual workloads — not generic provider blurbs
- [ ] A single named primary and a single named secondary provider are recommended (no hedging)
- [ ] Any hard compliance or geographic constraint that disqualifies a provider is flagged as overriding the score
Notes
- A provider can win the weighted math and still lose on a single hard constraint — always check disqualifying compliance/region gaps before finalizing.
- Existing enterprise agreements (especially Microsoft EAs) frequently move the effective-cost picture more than list-price comparisons suggest; weight commercial fit accordingly when those relationships exist.
- Multi-cloud is operationally expensive; recommend it only when a specific workload's advantage is large enough to justify the duplicated tooling, governance, and skills overhead.
- If inputs are sparse, score on documented provider capabilities and clearly label any assumption so the board can challenge it.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills