The Starting Point: Invisible to AI
In November 2025, the school approached Far & Wide for an AEO (Answer Engine Optimization) audit. The brand was nearly invisible to ChatGPT users searching for training options in its category.
The audit numbers
The school had a functioning website with course pages, a blog, and video content. None of it was structured in a way AI models could parse and use.
Why AI Models Ignored the Website
The technical audit found three root problems.
Broken site structure
The website ran on Framer, which auto-generated semantic HTML tags incorrectly. Headings were duplicated, applied to decorative elements (numbers, sums, icons), and used on regular paragraphs. The H1 was duplicated across the page and carried no informational value. For an LLM parsing HTML, this made the page structure unreadable.
Zero structured data
Google Rich Results Test returned "No items detected." No JSON-LD markup: no Organization schema, no BreadcrumbList, no FAQPage. AI models had no machine-readable way to understand what the business was, what it offered, or how pages related to each other.
Marketing language instead of facts
The copy used subjective claims like "leading academy" and "best training program." LLMs can't verify superlatives. Models default to sources with specific, structured information: numbers, tables, timelines, named entities.
What Was Done: 6 On-Site Changes
Far & Wide delivered the audit with prioritized recommendations in December 2025. The school's team implemented six on-site changes before the end of the month, all focused on making existing content readable and recommendable by AI models.
Semantic HTML cleanup
Every page got correct heading hierarchy: one H1 per page with a descriptive, keyword-relevant title, H2s for main sections, H3s for subsections. Decorative elements were removed from heading tags. Semantic containers (<main>, <section>, <article>, <header>) were added so LLMs could understand page layout.
Structured data implementation
JSON-LD markup was added across the site:
- Organization — brand name, description, links to social profiles via sameAs
- BreadcrumbList — site navigation path for every page
- FAQPage — structured Q&A blocks on key landing pages
Sitemap repair
The existing sitemap had a broken URL pointing to a non-existent page. The dead link was removed and replaced with the correct URL. A clean sitemap reduces crawl errors and ensures AI systems can discover all pages.
Author pages with Person schema
Individual pages were created for each trainer and program lead, marked up with schema.org/Person. Each page listed specialization, training format, and credentials. AI models could now attribute expertise to specific people, not just a brand name.
Content rewrite: from marketing copy to structured facts
All key pages were rewritten with one principle: replace subjective claims with structured, verifiable information.
- Superlatives and marketing language removed
- Comparison tables added (program vs. program, with entry requirements, workload, and expected outcomes)
- Content broken into self-contained sections where each block answers a specific user question
- Lists and tables used instead of narrative paragraphs for factual information
New "Terms and Finances" section
A dedicated page with structured pricing, entry requirements, progression criteria, and timelines, formatted as tables. This covered queries ("how much does it cost," "what are the requirements," "how long does it take") that previously had no structured answer on the site.
What Was NOT Done
Four recommended actions were not implemented during this period:
- Business profiles on Yandex Maps, Google Business, and 2GIS
- Review campaigns on industry forums
- YouTube channel restructuring (playlists, timestamps)
- Wikidata entity creation
The result came entirely from on-site changes. No external link building, no paid promotion, no social media campaigns.
The Result: 0 to 20 Leads in January 2026
Implementation took the rest of December. By January, the school recorded 20 inbound leads where prospects specifically mentioned ChatGPT as the source. Attribution method: a direct question during intake, "How did you hear about us?"
These weren't cold leads from advertising. Each person had been recommended the school by ChatGPT, had read about it in the response, and arrived already knowing about the programs and pricing.
Timeline
The school approached Far & Wide for an AEO audit
Audit delivered, recommendations implemented by end of month
20 leads attributed to ChatGPT
Why This Worked
An online school with existing content and an active website was invisible to AI recommendation engines. Not because the content was weak, but because the structure was broken. Six on-site changes, implemented in one month with no external signals work, moved the brand from zero AI-generated leads to 20 per month.
AI models recommend what they can read, parse, and verify. Content locked inside broken HTML, missing structured data, and wrapped in marketing language gets skipped. The competitor whose site is technically legible gets the recommendation instead.
Most competitors in training niches still optimize for Google, not for AI assistants. The ones who structure content for LLM readability first will capture the recommendations — and the leads that follow.
Want similar results?
Start with a Far & Wide AI Visibility Report to see where you stand and what to fix first.
Get your report