client-onboarding（客户入职）

Client Onboarding

Full onboarding playbook for a new client. Produces a Client Intelligence Package and Growth Strategy Recommendations.

Reference: docs/agency-playbook/client-launch-playbook.md for the full process documentation.

When to Use

• "Onboard [company name] as a new client"
• "Run the full intelligence gathering for [company]"
• "Create a growth strategy for [company]"

Prerequisites

• Company name and website URL
• Basic understanding of what the company does (or the agent will research it)

Phase 1: Intelligence Gathering (~30 min parallel agent work)

Setup

First, create the client folder structure:

clients/<client-name>/
├── context.md              # Will be populated during research
├── notes.md                # Running log
├── intelligence/           # All Phase 1 outputs go here
├── strategies/             # Will be populated after strategy approval
├── campaigns/
├── leads/
└── content/

Step 1: Company Deep Research

• Deep web research: product, pricing, team, funding, customers, tech stack, recent news
• Method: Web search + web fetch
• Output: clients/<client-name>/intelligence/company-research.md

Step 2: Competitor Deep Research

• Identify top 5-10 competitors, research positioning, pricing, strengths, weaknesses
• Method: Web search + web fetch
• Output: clients/<client-name>/intelligence/competitor-research.md

Step 3: Founder Deep Research

• Research founders: backgrounds, LinkedIn presence, thought leadership, public visibility
• Method: Web search + linkedin-profile-post-scraper
• Output: clients/<client-name>/intelligence/founder-research.md

Step 4: SEO Content Audit

• Full SEO footprint: content inventory, domain metrics, competitive gaps, brand voice
• Skill: seo-content-audit (orchestrates site-content-catalog + seo-domain-analyzer + brand-voice-extractor)
• Output: clients/<client-name>/intelligence/seo-content-audit.md

Step 5: AEO Visibility Check

• Test visibility across AI answer engines for key queries
• Skill: aeo-visibility
• Output: clients/<client-name>/intelligence/aeo-visibility.md

Step 6: Paid Ads Strategy Review

• Scrape active Meta and Google ads for client and top competitors
• Skill: meta-ad-scraper + google-ad-scraper
• Output: clients/<client-name>/intelligence/ad-strategy.md

Step 7: Industry Intelligence Scan

• Scan everything happening in the client's industry in the past week
• Skill: industry-scanner
• Output: clients/<client-name>/intelligence/industry-scan.md

Step 8: Current GTM Analysis

• Score the client's current GTM across all dimensions
• Skill: company-current-gtm-analysis
• Output: clients/<client-name>/intelligence/gtm-analysis.md

Parallel Execution Plan

Parallel Group A (general research — run simultaneously):
  Step 1: Company Deep Research
  Step 2: Competitor Deep Research
  Step 3: Founder Deep Research

Parallel Group B (automated audits — run simultaneously):
  Step 4: SEO Content Audit
  Step 5: AEO Visibility Check
  Step 6: Paid Ads Strategy Review
  Step 7: Industry Intelligence Scan
  Step 8: Current GTM Analysis

Groups A and B can run in parallel with each other.

Phase 2: Synthesis & Diagnosis

Read all Phase 1 outputs and synthesize into a single Client Intelligence Package.

Reference framework: docs/growth-frameworks.md

Diagnostic Steps

1. Assess PMF: Does the retention curve flatten? (Pre-PMF / PMF / Strong PMF)
2. Determine ACV tier: What viable channels does their price point support?
3. Identify growth motion: Product-led, marketing-led, sales-led, or blended?
4. Assess scaling stage: Pre-PMF → PMF → GTM Fit → Growth & Moat
5. Score current GTM: Rate each dimension A-F (from gtm-analysis output)
6. Map competitive landscape: Top 5 competitors with strengths/weaknesses
7. Identify opportunity gaps: Which Growth Matrix cells are empty?
8. Flag risk factors: Competitive threats, market risks, internal constraints

Output

File: clients/<client-name>/intelligence-package.md

Structure:

1. Company Profile
2. Stage Assessment (PMF, ACV, Motion, Scaling Stage)
3. Current GTM Scorecard (A-F per dimension)
4. Competitive Landscape
5. Industry Context
6. Opportunity Map (Growth Matrix gaps)
7. Risk Factors

Phase 3: Strategy Generation

Read the Intelligence Package and generate prioritized growth strategies.

Strategy Generation Process

For each identified opportunity gap:

1. Name the system: Map to the Growth Systems Taxonomy (Intelligence / Demand Creation / Pipeline)
2. Describe the gap: What's missing or broken?
3. Propose the solution: What system do we build? What skills power it?
4. Estimate impact: Expected lift based on available data
5. Sequence: P0 (immediate), P1 (4-6 weeks), P2 (8-10 weeks)
6. Score: ICE score (Impact x Confidence x Ease, each 1-10)
7. Tag the execution pattern: Add a structured <!-- execution... --> YAML block identifying the pattern, signal type, required skills, estimated cost, and estimated lead volume. See Structured Execution Tags below for the format.

Prioritization Rules

1. Activation before acquisition (if activation is broken, fix that first)
2. One channel deep before expanding
3. Engine over boost (compounding loops > one-time campaigns)
4. Always include at least one quick-win Pipeline strategy alongside longer-term Demand Creation
5. Match channel to ACV (no field sales for <$5K ACV)

Output

File: clients/<client-name>/growth-strategies.md

Format: P0/P1/P2 grouped strategies with gap, solution, tactical steps, expected impact, timeline.

See clients/vapi/growth-strategies.md as a reference example.

Structured Execution Tags

Every strategy in growth-strategies.md must include a machine-readable execution tag as an HTML comment block. This allows the client-packet-engine playbook to automatically route strategies to the correct skill chains.

Format

<!-- execution
pattern: signal-outbound
signal_type: job-posting
signal_keywords: ["DevOps", "SRE", "platform engineer"]
target_titles: ["VP Engineering", "CTO", "Head of Platform"]
estimated_leads: 50
estimated_cost: 0.80
skills_required:
  - job-posting-intent
  - company-contact-finder
  - email-drafting
-->

Fields

Pattern Selection Guide

Examples

Signal-Outbound (job posting intent):

### Strategy 1: DevOps Hiring Signal Outbound

Companies hiring DevOps/SRE roles likely need infrastructure tooling...

<!-- execution
pattern: signal-outbound
signal_type: job-posting
signal_keywords: ["DevOps", "SRE", "platform engineer", "infrastructure"]
target_titles: ["VP Engineering", "CTO", "Head of Platform"]
estimated_leads: 50
estimated_cost: 0.80
skills_required:
  - job-posting-intent
  - company-contact-finder
  - email-drafting
-->

Competitive-Displacement:

### Strategy 3: Capture Unhappy BigCo Customers

BigCo has declining review scores on G2 and recent feature removals...

<!-- execution
pattern: competitive-displacement
competitor_name: BigCo
target_titles: ["Head of Operations", "VP Product", "CTO"]
estimated_leads: 30
estimated_cost: 1.20
skills_required:
  - web-archive-scraper
  - review-scraper
  - company-contact-finder
  - email-drafting
  - content-asset-creator
-->

Manual (no automation available):

### Strategy 6: Partner Co-Marketing Program

Build joint content and referral agreements with complementary tools...

<!-- execution
pattern: manual
estimated_leads: 0
estimated_cost: 0
skills_required: []
-->

Rules

1. Every strategy gets tagged — even manual ones. This ensures the packet engine can account for all strategies.
2. Be specific with signal_type — don't use generic descriptions. Map to the exact signal source the skill will scan.
3. List all skills in chain — in execution order. The packet engine uses this to plan parallel execution.
4. Estimate conservatively — overestimate cost, underestimate leads. Better to over-deliver than under-deliver.
5. Use manual sparingly — if a strategy can be even partially automated, tag it with the automatable pattern and note limitations in the strategy description.

Human Checkpoints

• After Phase 2: Review the Intelligence Package for accuracy before generating strategies
• After Phase 3: Review strategies with client before implementation