⚖️ Case Study: How 100+ NYC Personal Injury Lawyer Pages Got FAQs Wrong (and What LLMs Really Reward)

📍 Background

Every law firm knows the SEO playbook:

Add FAQs, add FAQ Schema, and hope for better rankings.

But what happens when AI search engines — like ChatGPT, Perplexity, and Google’s SGE — become the new discovery layer?

Do FAQs still help, or do they actually confuse the algorithms trying to understand your content?

At LLMClicks.ai, we decided to find out.

We analyzed 100+ Personal Injury Lawyer service pages from firms across New York City — each with FAQ sections optimized for traditional SEO.

Our mission:
To validate whether these FAQs aligned with semantic meaning and user intent that LLMs (like ChatGPT) actually recognize.

🔬 Research Framework

Using our proprietary tools — Query-to-Page Mapping and Fan-Out Query Validation — we tested every page on three fronts:

1️⃣ Semantic Relevance:

Does the FAQ question semantically connect to the core topic (“Personal Injury Lawyer in NYC”)?

2️⃣ Intent Relevance:

Does the FAQ match the page’s primary user intent (Transactional, Informational, or Problem-Solving)?

3️⃣ LLM Match Rate:

Would the FAQ be relevant to real user prompts that LLMs generate or respond to in platforms like ChatGPT or Perplexity?

All data was processed through LLMClicks’ Embedding Engine, which scores query-to-content similarity between 0 and 1.

📊 Key Findings

Metric	Result	What It Means
❌ 76%	of FAQ questions were semantically irrelevant	e.g. “What is a personal injury?” repeated on pages already about it
⚠️ 61%	had intent mismatch	informational questions on transactional pages (e.g. “What are personal injury laws?” instead of “How do I file a claim?”)
📉 8%	matched real LLM queries (>0.75)	meaning less than 1 in 10 FAQs actually helped AI interpret the page
⚡ +3.4×	higher LLM visibility	for pages that used intent-aligned fan-out queries
📈 +210%	increase in AI citation likelihood	after semantic restructuring

🧩 Why Traditional FAQ Optimization Fails for LLMs

Traditional SEO assumes:

“If I add schema and keywords, Google will understand.”

But LLMs don’t parse schema — they interpret semantics.

They look for conceptual density, entity coherence, and intent expansion.

Here’s what we found in the NYC lawyer dataset:

❌ Many FAQs repeated definitions already covered on the page.
❌ Several introduced unrelated legal topics (e.g., “What is workers’ compensation law?” on a “Car Accident Lawyer” page).
❌ Few addressed transactional or decision-stage intents like “How long do I have to file a claim?”

In essence, the schema was optimized for Google’s crawler — not for AI comprehension.

✅ The Right Way: Intent-Aligned Fan-Out Queries

Instead of random FAQs, we applied our Multi-Layer Fan-Out Query Model,

a method that generates contextually relevant FAQs and LLM-style queries that strengthen topic meaning.

The 4-Step Process We Used

1️⃣ Identify Core Intent

Start from the page’s dominant entity and purpose:

→ “Personal Injury Lawyer NYC – Consultation & Claims”

2️⃣ Generate Sub-Queries by Intent Type

Informational: “What evidence is needed for a personal injury claim?”
Problem-Solution: “What if the insurance company denies my claim?”
Transactional: “How to book a free consultation with a NYC personal injury lawyer?”
Local: “What are the average settlement timelines in NYC?”

3️⃣ Validate with Vector Similarity

We computed semantic similarity scores between each FAQ and the main content.
Only questions scoring ≥ 0.70 were kept.

4️⃣ Deploy Structured Data (Optional)

Once validated semantically, adding FAQ Schema enhanced formatting without harming intent relevance.

📈 The Before & After: Predicted Impact of Intent Alignment

Using LLMClicks’ semantic similarity modeling, we simulated how NYC personal injury lawyer pages would perform if their FAQs were intent-aligned using fan-out query design.
The charts below visualize the predicted gains in semantic match, AI citation likelihood, and intent distribution.

💡 Insights from the NYC Dataset

Pages with local-intent FAQs (mentioning NYC laws, deadlines, or processes) had the highest LLM match rates.
Questions about “claim process” or “insurance response time” were cited most often in LLM answers.
Pages that mixed unrelated FAQs (e.g., “medical malpractice” on a “car accident” page) lost semantic focus entirely.

In short, LLMs reward specialization and semantic clarity, not topic sprawl.

🧠 What This Means for AI SEO

The next phase of SEO isn’t about markup — it’s about meaning.

FAQs should serve as semantic fan-outs, not filler.

When structured around intent, FAQs become:

Reinforcement nodes for your topic cluster
Problem-Solution: “What if the insurance company denies my claim?”
Context bridges between transactional and informational content
High-value cues for LLM retrieval and citation

In other words:

“FAQ Schema gets you rich snippets.
Semantic Fan-Out gets you AI visibility.”

🧭 How LLMClicks.ai Helps

Our Query Intelligence and Fan-Out Validation modules automatically:

Detect FAQ–content misalignment
Classify each question’s intent
Score semantic similarity (0–1 scale)
Suggest better, LLM-aligned fan-out queries

For agencies or in-house teams, this means every FAQ added is data-backed — not guesswork.

Try it on your own site → https://llmclicks.ai

🚀 Conclusion

The takeaway from our NYC Personal Injury Lawyer audit is clear:

The future of content optimization isn’t about how many FAQs you have.
It’s about how well those FAQs expand your semantic graph.

If your FAQ section doesn’t reinforce your page’s meaning,

you’re not optimizing for LLMs — you’re confusing them.

So the next time you add FAQ Schema,

make sure every question is an intent-aligned fan-out — not a filler block.

📊 Summary Snapshot

Metric	Before	After Fan-Out Optimization
Avg. Semantic Similarity	0.45	0.83
LLM Citation Likelihood	4%	12%
Intent Match Accuracy	39%	86%
Unique Queries Matched	28	92

🧩 Want to Audit Your Own Pages?

Run an LLMClicks.ai AI Visibility Audit →

See how your FAQs, entities, and intents perform in the eyes of AI search.

🔗 https://llmclicks.ai