aos-analyze-gtm

Data access

This skill's data lives in the granted folder — the folder Cowork was given access to, which is one client's folder (no per-client nesting). The granted-folder root is the working directory. Resolve zones (client/, brand/, inbox/, ontology/, deliverables/, …) per docs/data-access-router.md and the AOS_CONFIG.md manifest at the granted-folder root. Never hard-code paths beyond the documented zone layout. Client identity (the client name / slug) is read from client/CLIENT_CONFIG.md and the client field of AOS_CONFIG.md — it is never a directory level. Bash + filesystem on the granted folder is the contract; the router is an optimization.

Language

Resolve communication-language and content-language from AOS_CONFIG.md during context assembly (per docs/language-context.md) — never hard-code a language. Talk to the user in communication-language; write client-facing artifacts in content-language.

Purpose

Analyzes a business's Go-To-Market strategy against the complete GTM Strategist framework to identify gaps, weaknesses, and opportunities. Provides a clear diagnosis of what is missing or underperforming before action planning.

The core principle:

"A go-to-market strategy is the difference between a hope and a plan." — J.P. Eggers, NYU Stern School of Business

Posture: Discovery, not pronouncement. Present observations with questions, not conclusions. Show calculations. Invite disagreement. End every deliverable with "What did we get wrong? What's missing?"

Diagnostic guardrails

• Multi-domain. If client/CLIENT_CONFIG.md (or client/DOMAIN_CHANNEL_MAP.yaml) lists 2+ domains, load the domain-channel map FIRST. Filter all channel/ROAS/spend queries by domain. Never use account-level totals as a single domain's metric.
• Evidence classification. Every evidence item must be tagged: [DATA], [OBSERVED], [STATED], [NARRATIVE], [INFERRED], [HEARSAY]. Only [DATA] and [OBSERVED] items can carry a causal claim. Keep data confidence separate from causal confidence. GA4 alone is insufficient for causal diagnosis — always check platform-side data (Meta Ads Manager, Google Ads, etc.).
• Attribution windows. Before comparing conversion metrics across platforms, account for window mismatch — Google Ads (30-day click) ≠ Meta (7-day click) ≠ GA4 (data-driven). Never sum platform conversions; use GA4 as the cross-platform arbitrator. Flag any window mismatch in the finding.
• Currency normalization. Before summing monetary values across domains/markets, convert to the client's reporting currency and state the conversion rate and its source.
• Dedup. Before emitting new RECs, check ontology/recommendations/ for existing RECs on the same gap.

Trigger

Use this skill when:

• The client is launching a new product/service
• The current GTM isn't producing results
• The client feels "stuck" in growth
• Before creating a GTM action plan
• The client is unsure where to focus GTM efforts
• Pivoting to a new market or segment

The GTM framework — the 6 core decisions (+ system)

┌─────────────────────────────────────────────────────────────────────────┐
│                         GTM STRATEGIST FRAMEWORK                         │
├─────────────────────────────────────────────────────────────────────────┤
│  1. MARKET          "What terrain should we win?"                        │
│  2. EARLY CUSTOMER  "Who do we serve FIRST?"                             │
│  3. PRODUCT         "What value do we deliver?"                          │
│  4. PRICING         "How do we capture value?"                           │
│  5. POSITIONING     "How do we stand out?"                               │
│  6. GROWTH          "How do we acquire customers?"                       │
│  + GTM SYSTEM       "How do we execute?"                                 │
└─────────────────────────────────────────────────────────────────────────┘

Process

Step 0 — Preflight

1. Confirm the working directory is the granted-folder root. Read AOS_CONFIG.md for the zone manifest and client identity.
2. Verify client/CLIENT_CONFIG.md exists. If not, suggest running aos-onboard.
3. If the client has 2+ domains, load client/DOMAIN_CHANNEL_MAP.yaml before any analysis.
4. Scan the brand/ zone for existing intelligence (POSITIONING, ICP, 7LAYER_DIAGNOSTIC) and inbox/ for discovery material.

Step 1 — Gap analysis (7 areas)

Work each of the 7 areas — MARKET, EARLY CUSTOMER PROFILE, PRODUCT, PRICING, POSITIONING, GROWTH, GTM SYSTEM. For each, ask its key questions, look for gap indicators, and compare against "what good looks like." The full per-area question lists, symptom→gap tables, and "what good looks like" criteria live in reference/gap-analysis-framework.md.

Step 2 — Rate and prioritize

Rate each area 🔴 Critical / 🟡 Gaps Present / 🟢 Solid. Build the Gap Priority Matrix (severity × urgency × effort) and pick the top 3 gaps with a recommended first focus.

Step 3 — Belief check

Surface limiting beliefs that may be causing the gaps — what would someone have to believe to get this result? Trace deeper if the GTM block is identity-level (consider running aos-diagnose-7layer for L0 work).

Step 4 — Format and write

Format the deliverable using the output template in reference/gap-analysis-framework.md. Save it under deliverables/<YYYY-MM>/ — resolve the deliverables zone via AOS_CONFIG.md, never hard-code the path.

Key distinctions

• ECP (Early Customer Profile) — for GTM: who to target FIRST.
• ICP (Ideal Customer Profile) — for Growth: who to target at SCALE.
• Beachhead Strategy — focus on one segment before expanding.
• GTM has a 3–18 month lifespan — urgency matters. Iterate fast, learn faster.

Output sections

Minimum content for every delivery:

• Executive summary — overall GTM health + top 3 gaps + recommended focus
• Detailed gap analysis — all 7 areas
• Gap priority matrix
• Belief check
• Recommended next steps
• What did we get wrong? What's missing?

Provenance

Every artifact this skill writes carries the standard provenance block in its frontmatter — see docs/artifact-versioning.md §1. Stamp all four fields:

generated_by: <this skill's name>      # the name: frontmatter value
skill_version: <this skill's version>  # the version: frontmatter value
generated_date: <YYYY-MM-DD>           # the date written
aos_schema: <schema-version>           # read from AOS_CONFIG.md

Add it to whatever domain frontmatter the artifact already carries; never hard-code skill_version or aos_schema — read them at write time. This is what lets a granted folder be migrated when the plugin or schema changes.

Hard Rules

1. This skill is for analysis — not action planning. Diagnose gaps; do not write the plan.
2. Never sum platform conversions across mismatched attribution windows.
3. Multi-domain: filter every channel/spend/ROAS metric by domain; never use account-level totals.
4. Cite evidence with its class; a single channel's analytics is never a causal claim on its own.
5. Single client — operate only within the granted folder; never reach outside it.

Integration

• Upstream: aos-onboard (scaffolds the granted folder); aos-diagnose-7layer (the L0–L7 diagnosis — run first when the GTM block looks identity-level).
• Downstream: GTM action planning (close the gaps); aos-build-brand to fix a positioning gap; idea/assumption validation when the gap is market/ECP; aos-build-brand-system consumes GTM findings into the Client Intelligence Profile.

Key quotes

"Over 95% of businesses fail in their first three years of existence."

"If you are serving everybody, you are not doing a fantastic job for anybody."

"A winning strategy is to create a critical mass of activities for success."

"The best GTM strategies are simple yet proprietary."

"In GTM, you need to be extremely careful to resist shiny objects and commit your limited resources to the mission."

What did we get wrong? What's missing?