Understanding AI logic is the first step to becoming discoverable. Here's what happens when an attorney asks for expert recommendations.
AI Doesn't Think Like Humans
When an attorney asks a colleague for expert recommendations, that colleague considers reputation, personal experience, and industry knowledge. The recommendation is inherently subjective.
AI systems work differently. They parse, categorize, and rank based on structured patterns and verifiable signals. Understanding this difference is essential to becoming discoverable.
The AI Evaluation Framework
When ChatGPT, Perplexity, or similar AI systems encounter content about an expert witness, they evaluate it across four primary dimensions:
1. Relevance: "What Is This About?"
AI first determines whether content is relevant to the query. It looks for:
Clear Topic Signals
- Headlines and headings that state expertise explicitly
- Structured data markup indicating professional category
- Natural language descriptions using expected terminology
Specificity
- Generic terms like "medical expert" are less useful than "pediatric neurologist specializing in birth injury"
- AI prefers precise expertise descriptions it can match to specific queries
What This Means for You:
Write your expertise descriptions in clear, specific language. Use the terms attorneys would actually search for. Avoid jargon that sounds impressive but isn't searchable.
2. Credibility: "Is This Trustworthy?"
Relevance alone isn't enough. AI must assess whether to trust the information:
Verifiable Credentials
- Links to actual publications on recognized platforms
- Institutional affiliations that can be confirmed
- Professional certifications from known organizations
Source Authority
- Information on authoritative domains carries more weight
- University affiliations, major publications, recognized organizations
- Consistent information across multiple reputable sources
What This Means for You:
Don't just list credentials—link to verification. A publication listed without a link is less valuable than one with a working URL to the actual article.
3. Recency: "Is This Current?"
AI systems prioritize fresh, maintained content:
Update Signals
- When was the page last modified?
- Are there recent dates mentioned (publications, cases, activities)?
- Does the content reference current events or recent developments?
What This Means for You:
Update your profiles regularly, even if only minor changes. Add new credentials as they occur. A profile touched last month outranks one untouched for two years.
4. Quotability: "Can I Use This?"
AI systems that generate recommendations need content they can reference:
Clear, Quotable Statements
- "Forensic accountant with 15 years investigating securities fraud" is quotable
- "Experienced financial professional" is not useful
What This Means for You:
Write your bio as if it will be quoted directly. Use specific, factual statements. Include numbers: years of experience, case counts, publication numbers.
How the Dimensions Interact
AI doesn't evaluate these factors in isolation—they interact:
High Relevance + Low Credibility = Weak Recommendation
The expert matches the query but lacks verification. AI may mention them with caveats or skip entirely.
High Credibility + Low Relevance = No Recommendation
The expert is clearly qualified but not for this specific query. Specificity matters.
High Relevance + High Credibility + Low Recency = Possible Recommendation
Strong credentials but potentially outdated. AI may recommend but note the uncertainty.
High Across All Four = Strong Recommendation
The expert appears in results with confidence and detail.
Practical Example
Consider how AI handles two different expert profiles for the same query: "forensic accountant expert witness securities fraud California"
Expert A:
- LinkedIn profile says "Financial consultant" (relevance: low)
- No links to publications (credibility: low)
- Profile last updated 2022 (recency: low)
- Bio: "Experienced professional helping clients succeed" (quotability: low)
Result: Not recommended
Expert B:
- Profile says "Forensic accountant specializing in securities fraud investigations" (relevance: high)
- Links to two articles in Journal of Forensic Economics (credibility: high)
- Profile updated last month with new case type added (recency: high)
- Bio: "CPA/CFE with 18 years investigating insider trading, Ponzi schemes, and SEC violations in California state and federal courts" (quotability: high)
Result: Recommended with confidence
Same underlying qualifications, radically different AI outcomes.
What AI Can't Evaluate
Understanding limitations helps set realistic expectations:
- AI can't verify credentials independently (only check for consistency and links)
- AI can't assess actual testimony quality or courtroom presence
- AI can't evaluate personality fit with specific attorneys
- AI can't guarantee accuracy of self-reported information
AI provides discovery and initial filtering—not comprehensive vetting. This is actually an advantage: it means you can influence discoverability through optimized presentation of genuine credentials.
Applying This Understanding
The rest of this guide builds on these four evaluation dimensions:
- Structured data optimization addresses relevance and quotability
- Verifiable credential links address credibility
- Regular updates and fresh content address recency
- Profile optimization addresses all four simultaneously
Understanding how AI thinks is the foundation. Now let's optimize for it.
Frequently Asked Questions
Does AI just copy what's on the first search result?
No. AI systems synthesize information from multiple sources and evaluate credibility. Being on one page isn't enough—you need consistent, verifiable presence across sources.
Can I optimize for one AI system specifically?
Most AI systems (ChatGPT, Perplexity, Claude) use similar underlying logic. Optimizing for structured data and credibility signals works across platforms.
How often do AI systems update their knowledge?
It varies. ChatGPT's browsing feature accesses current web content. Other systems update their knowledge bases periodically. Consistent optimization ensures you're ready whenever they crawl.
