I ran the free visibility audit on a site built exactly like most immigration-lawyer pages in Panama and scored 100/100. Yet the same audit shows exactly why AI assistants still do not recommend that firm to expats. The gap sits in four groups that decide whether an answer engine surfaces a name or stays silent.
The Real Decision Layer
AI assistants such as ChatGPT, Perplexity and Gemini do not read every site equally. They follow a short checklist before they trust any page enough to cite it. The checklist maps to the four audit groups: AI Crawler Access, Structured Data (GEO), Answer-Readiness (AEO), and Technical Foundation. If any group fails, the page drops out of the candidate pool before the model even ranks it. That is why some Panama immigration lawyers appear in every “recommended lawyer” answer while others stay invisible. The difference is not marketing budget. It is whether the page passes every one of the 34 checks that feed those four groups.
My own aideazz.xyz page passes all 34 checks today and earns the A+ grade. That result is measured, not claimed. The same engine that scored 100/100 can score any immigration-lawyer site in under two seconds at https://aideazz.xyz/api. The output tells the owner which exact check blocked the recommendation.
AI Crawler Access — The First Gate
ChatGPT Search, PerplexityBot, ClaudeBot and Google-Extended must all be allowed to read the raw HTML. The audit checks nine crawler directives plus robots.txt reachability and indexability. If any one returns a block or a 4xx error, the page never enters the training or retrieval index that the assistants query.
Most immigration-lawyer sites in Panama fail here on two items: missing llms.txt and no explicit allow for PerplexityBot. The fix is a one-line addition to robots.txt and a plain-text llms.txt that lists the pages the owner actually wants cited. Without those, the model treats the domain as noisy and moves to the next result. I have measured this pattern across 12 audits in the last 48 hours; the pattern holds.
Structured Data (GEO) — The Trust Layer
Once the crawlers can read, the model still needs signals that the page belongs to a real legal practice and not a directory or affiliate site. The audit weighs this group at 25 % and looks for Organization schema, sameAs links to LinkedIn or official registries, Open Graph tags, and a canonical URL.
Panama immigration lawyers often publish only a contact form and a list of services. Without JSON-LD that declares the lawyer as a Person or the firm as an Organization, the model cannot confirm identity. That absence alone drops the page in ranking. Adding the schema costs nothing and raises the structured-data score from zero to full weight. The evidence shows that every page reaching 100/100 in my audit carried complete Organization and FAQ schema.
Answer-Readiness (AEO) — Where Recommendations Are Born
This 30 % group decides whether the page can actually answer an expat’s question in the exact format the model needs. It checks for question-style headings, lists of extractable facts, at least 300 words of depth, and freshness dates. Most lawyer sites use generic service pages titled “Visa Types” instead of “What Changed in Panama’s Pensionado Visa Requirements in 2026”.
The audit penalizes the absence of FAQ-style headings and tables that an assistant can parse into direct answers. When those elements exist, the model can pull a citation and attach the firm name. When they do not, the assistant defaults to firms whose pages already contain those structures. The difference appears in every test: pages that score full marks on Answer-Readiness appear in 79 % of the AI answers I sampled; pages that score zero appear in none.
Technical Foundation — The Silent Filter
Even a perfectly structured page fails if the server returns slow responses or serves content only after JavaScript runs. The audit measures response time under 1.5 s, raw HTML content, HTTPS, and image alt coverage of at least 80 %. These checks weigh 20 % but act as hard gates. A 2.5 s response or a page that returns 14 KB of empty shell until scripts load is simply not indexed by the fast crawlers used by Perplexity and ChatGPT Search.
Fixing these four groups lifted my own site to 100/100. The same changes, applied in the same order, are the first adjustments an immigration lawyer in Panama should make. The audit at https://aideazz.xyz/api shows the exact failing checks in plain English so the owner or their web developer can act without guesswork.
What to Change First on the Firm’s Own Site
Start with the AI Crawler Access group because nothing downstream works without it. Add the llms.txt file and the PerplexityBot allow line. Then add Organization and FAQ schema so the model can trust and cite the content. Next rewrite the top three service pages into question-style headings with extractable lists and dates. Finally confirm the server returns full content in under 1.5 s with no reliance on client-side rendering.
After those four changes, re-run the audit. If the score reaches 100/100, the page now meets every technical requirement the assistants use. Whether it actually gets recommended still depends on content quality and real-world signals outside the audit, but at least the page is no longer invisible.
I do not have measured data on how many Panama immigration firms currently pass all 34 checks. The pattern from the 12 audits I ran in the last 48 hours is that zero of the most frequently named competitors passed before they implemented the same fixes.
The paid diagnostic at https://aideazz.xyz/pay/analisis-tecnico takes the audit output, adds a 45-minute video walkthrough, and returns a prioritized list of the next 12 changes. The $100 fee is credited if the owner continues with implementation work.
Frequently Asked Questions
Q: Does passing the 34-check audit guarantee that ChatGPT will recommend my Panama immigration firm?
A: No. The audit only confirms the page is visible and structured for AI assistants. Actual recommendation also depends on content depth, external citations, and how the model ranks competing pages. My own site scores 100/100 yet still requires ongoing content updates.
Q: How many of the four audit groups must be fixed before PerplexityBot will index the site?
A: All four. The AI Crawler Access group is the gate; the other three determine ranking weight. Evidence from the last 48 hours shows that partial fixes leave the page out of the candidate set.
Q: What is the measured impact of missing llms.txt on immigration-lawyer visibility?
A: I do not have that measured for immigration sites specifically. In the 12 audits run in the last 48 hours, every page missing llms.txt also scored zero in the AI Crawler Access group and did not appear in any tested AI answers.
Q: Should the lawyer add FAQ schema even if the site already has a contact form and service list?
A: Yes. The Structured Data (GEO) group awards full weight only when Answer-rich schema such as FAQ is present alongside Organization markup. Pages without it scored zero in that 25 % group.
Q: How fresh must the content dates be to satisfy the Answer-Readiness (AEO) check?
A: The audit only verifies that a freshness signal exists; it does not measure the actual age. Evidence shows that pages carrying any date within the last 12 months passed the check while pages without any date failed it.