Our support content gets zero AI citations - what are we doing wrong?

I’ve audited hundreds of support sites for AI visibility. Here’s what I typically find:

Why support content often fails with AI:

The fix framework:

1. Make content extractable Each section should stand alone. If AI lifts one paragraph, it should make complete sense.

2. Lead with answers Don’t build up to the answer. State it first, then explain.

3. Use explicit Q&A format Real questions as headings, complete answers below.

4. Add TL;DR sections 50-100 word summary at top of each article.

**Your 500 articles aren’t worthless - they need restructuring, not rewriting.

Structure patterns that work for AI:

The “Answer First” template:

[Clear question as H2]

[1-2 sentence direct answer]

[Detailed explanation]

[Step-by-step if applicable]

[Related FAQs]

Example transformation:

Before: “When working with our system, you might encounter situations where data needs to be exported. There are several ways to do this depending on your needs. Let’s explore the options available to users on different plans…”

After: “How do I export my data?

Export data by going to Settings > Data > Export. Select CSV or JSON format and click Download.

Step-by-step:

1. Navigate to Settings (gear icon)
2. Click Data in the left menu
3. Select Export
4. Choose format: CSV or JSON
5. Click Download

Note: Enterprise users have additional export options including API access and scheduled exports.”

The difference:

• Question is explicit
• Answer comes first
• Steps are numbered
• Variations are noted

Technical implementation for support content:

Schema markup for support articles:

{
  "@type": "HowTo",
  "name": "How to export data",
  "step": [
    {
      "@type": "HowToStep",
      "name": "Open Settings",
      "text": "Click the gear icon in the top-right corner"
    }
  ]
}

For FAQ sections:

{
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How do I export data?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Export data by going to Settings > Data > Export..."
      }
    }
  ]
}

Don’t hide content:

AI crawlers may not render:

• JavaScript-loaded content
• Tab panels
• Accordions
• Modal popups

Keep critical information in the main HTML flow.

robots.txt check:

Ensure AI crawlers aren’t blocked:

• GPTBot
• PerplexityBot
• ClaudeBot
• Google-Extended

Prioritization framework for 500+ articles:

Tier 1: High-priority optimization (50 articles)

Tier 2: Medium priority (100 articles)

• Product core features
• Integration guides
• Setup/onboarding content

Tier 3: Low priority (rest)

• Edge cases
• Legacy features
• Rarely asked questions

How I’d approach your 500:

Week 1-2: Identify and prioritize

• Pull analytics on top articles
• Map to common support queries
• Test current AI visibility

Week 3-6: Optimize Tier 1

• Restructure top 50 articles
• Add schema markup
• Create FAQ sections

Week 7-10: Tier 2 optimization

• Apply learnings from Tier 1
• Focus on high-value features

Metric to track:

For each optimized article:

• Test in ChatGPT/Perplexity before
• Test again 4-6 weeks after
• Document citation improvements

Common support content mistakes for AI:

Mistake 1: PDF-only documentation

PDFs lack:

• Structured headings
• Semantic markup
• Schema support
• Easy crawling

Fix: HTML versions of critical docs.

Mistake 2: Video without text

AI can’t watch videos. Must have:

• Full transcripts
• Text summaries
• Step-by-step text alternatives

Mistake 3: Images without descriptions

“See screenshot below” is useless for AI.

Include: “The Settings menu is in the top-right corner. Click the gear icon to see Account, Security, and Export options.”

Mistake 4: Internal jargon

Bad: “Navigate to the CRM module and configure the webhook endpoint.”

Good: “Go to the Customer Management section (CRM) and set up the automatic notification URL (webhook) that sends data to your other systems.”

Mistake 5: Assuming user knowledge

Don’t assume users know where things are. Be explicit about locations and paths.

Real results from support content optimization:

Our before/after data:

What moved the needle most:

1. Adding TL;DR sections - 15 minutes per article, biggest impact
2. Converting to FAQ format - For troubleshooting articles
3. Schema markup - HowTo and FAQ schema
4. Removing accordions - Put all content in main flow

Unexpected benefit:

Customers started saying “ChatGPT told me about your feature.” This led to feature discovery they wouldn’t have found otherwise.

Process we used:

• 1 technical writer, 20 hours/week
• 60-day sprint on top 50 articles
• Monthly maintenance for new content

Am I Cited helped us track which articles were getting picked up and which still needed work.

Writing guidelines for AI-friendly support content:

Language rules:

1. Short sentences - Under 25 words
2. Active voice - “Click Save” not “Save should be clicked”
3. Define jargon first use - “API (Application Programming Interface)”
4. Specific locations - “top-right corner” not “at the top”

Structure rules:

1. One topic per section - Don’t mix features
2. Numbered steps - For any process
3. Bullet lists - For options/features
4. Tables - For comparisons

Answer completeness test:

Ask yourself: “If this paragraph was the ONLY thing someone read, would they have a complete answer?”

If no, add more context.

Example test:

“Enable two-factor authentication in your account settings.”

Incomplete. Better:

“Enable two-factor authentication by going to Settings > Security > Two-Factor Auth. Click Enable and follow the prompts to connect your authenticator app. You’ll receive a verification code to confirm setup.”

This is exactly what I needed. Here’s my action plan:

Week 1: Audit & Prioritize

• Pull top 50 articles by traffic
• Cross-reference with common support queries
• Test current AI visibility baseline
• Identify articles with hidden/accordion content

Week 2-4: Tier 1 Restructure

• Add TL;DR to top 25 articles
• Convert to question-answer format
• Remove context-dependent language
• Add detailed image descriptions

Week 5-6: Technical Implementation

• Add FAQ and HowTo schema
• Check robots.txt for AI crawlers
• Convert critical accordions to visible content
• Create HTML versions of PDF guides

Week 7-8: Tier 2 Start

• Apply template to next 50 articles
• Train team on new writing guidelines

Metrics:

• Baseline AI citations before
• Track weekly with Am I Cited
• Goal: 20%+ citation rate in 90 days

Key principles:

• Answer first, explain second
• Every paragraph stands alone
• No assumed context
• Explicit locations and actions

Thanks everyone - this completely changes how I think about support content.