plg-ai-funnel（PLGAI 漏斗）

PLG AI Funnel: Product-Led Growth in the Agent Era

The Paradigm Shift

Old PLG Funnel:

Landing Page → Free Trial → Activation → Conversion

New PLG Funnel:

Agent Query → Documentation Scan → Feature Match → Recommendation

The buyer's first interaction is no longer your landing page—it's an AI agent scanning your documentation to answer their question.

The Four Stages

Stage 1: Agent Query

What happens: User asks AI "What tool can help me [problem]?"

Optimization goals:

• Brand appears in AI's consideration set
• Correct category association
• Problem-solution mapping exists in AI's knowledge

Tactics:

Audit questions:

• Does AI know your brand when asked directly?
• Does AI associate your brand with your category?
• Do competitors appear but you don't?

Tool: entity-builder agent for authority building

Stage 2: Documentation Scan

What happens: AI scans your docs, help center, marketing pages to understand capabilities.

Optimization goals:

• Content is AI-extractable (chunked, structured)
• Answers are front-loaded (not buried)
• Each page passes the "Taco Bell Test" (stands alone)

Tactics:

The Extractability Checklist:

☐ First sentence directly answers the page's implied question
☐ H2/H3 headers are questions or clear topic labels
☐ Tables used for comparisons and feature lists
☐ Each section makes sense without surrounding context
☐ No "as mentioned above" or "see below" dependencies

Tool: llm-optimizer agent for content optimization

Stage 3: Feature Match

What happens: AI matches user's specific needs to your product's capabilities.

Optimization goals:

• Features described in user-problem terms
• Use cases explicitly mapped to capabilities
• Limitations clearly stated (builds trust)

Tactics:

Feature Documentation Template:

## [Feature Name]

**Problem it solves:** [User problem in their words]

**How it works:** [1-2 sentence explanation]

**Best for:** [Specific use cases]

**Limitations:** [What it doesn't do]

**Example:** [Concrete scenario]

Anti-pattern: Feature pages that describe functionality without connecting to user problems.

Stage 4: Recommendation

What happens: AI decides whether to recommend your product and how to position it.

Optimization goals:

• Clear differentiation from alternatives
• Social proof AI can cite
• Product tie-backs throughout content

Tactics:

The Product Tie-Back Rule: Every 1-2 paragraphs of educational content should include how your product relates.

• ❌ "Lead scoring helps prioritize prospects"
• ✅ "Lead scoring helps prioritize prospects—[Product] automates this with AI-powered scoring"

Tool: aeo-scorecard skill for measuring recommendation success

PLG × AEO Integration

Quick Audit Workflow

1. Test 10 queries your buyers ask
   → Does your brand appear? (Stage 1)

2. Check if AI cites YOUR content
   → Or competitor/third-party? (Stage 2)

3. Ask AI about specific features
   → Does it know your capabilities? (Stage 3)

4. Ask "Should I use [Product] for [use case]?"
   → What's the recommendation? (Stage 4)

Common PLG AI Gaps

Related Tools

• llm-optimizer - Deep content optimization for Stage 2
• entity-builder - Authority building for Stage 1
• aeo-scorecard - Metrics framework for all stages
• /aeo-workflow - Full implementation workflow
• query-expansion-strategy - Understanding query fan-out