Here is the short version: a focused education brand with no massive domain authority earned more AI citations than one of the world's biggest professional-services firms — for the same category of training. Not because it produced more content. Because it produced the right kind of content. This post explains exactly what that means and how to replicate it.
TL;DR
- FIT Institute (fitiedu.com) was cited by Google's AI Overview across approximately 8 course lines spanning education, taxation, and fashion.
- For tax-agent training, Google's AI Overview cited fitiedu.com ahead of PwC's Academy. ChatGPT cited it for the same queries.
- The result came from answer-shaped content, catalog-level topic authority, and E-E-A-T signals legible to AI — not from technical tricks or a high domain authority score.
- llms.txt and schema were not yet implemented at time of capture. They are upside, not prerequisite.
- Full case study with screenshots: fitiedu-geo case study
The verified result: what actually happened with FIT Institute
FIT Institute (fitiedu.com) is an education brand I work with remotely. The result documented in the full case study — including screenshots — was captured on 2026-06-17:
- Google's AI Overview cited fitiedu.com across approximately 8 course lines.
- Those citations spanned three distinct industries: education, taxation, and fashion.
- For tax-agent training searches, FIT Institute was cited by Google's AI Overview ahead of PwC's Academy.
- ChatGPT independently cited fitiedu.com for the same category of queries.
I state the caveat plainly: AI Overview content varies by user, device, location, and query date. These results were captured from a specific context and have not been independently re-validated from a neutral IP. Stating that caveat is not a weakness — it is exactly the kind of honest, verifiable framing that makes a claim worth citing in the first place.
What matters is the mechanism. The citations were not luck. They came from a specific, repeatable method.
Why out-citing a big brand matters more than out-ranking it
A page-one ranking is visible to people who scroll. An AI citation is visible to people who do not. When Google's AI Overview answers a query, it generates a response and names a handful of sources. Those sources get the attention. The organic results below the fold get the scroll.
For the tax-agent training category, the AI Overview was summarising training options and naming providers. FIT Institute was named. PwC's Academy — with its global brand recognition and presumably higher domain authority — was not the one cited. The user's first impression of "who provides this training" was shaped by fitiedu.com, not by PwC.
This is why generative engine optimization is now a distinct goal from SEO. You can rank on page one and still never get mentioned. Conversely, a focused, well-structured source can be cited even when a larger competitor has more raw authority. The AI is not ranking pages — it is selecting the most citable passage for a given question.
The GEO method: how FIT Institute earned AI citations
1. Answer-shaped content at catalog scale
Every course page at FIT Institute was written to answer the question a prospective student would type, not to describe the course from the institution's perspective. The opening of each page states what the course is, who it is for, and what outcome it delivers — in plain language, in the first paragraph.
AI engines extract answers from content that puts the answer at the top. Content that builds up to its point over several paragraphs rarely gets extracted, because the engine cannot be confident the first sentence is actually the answer. Answer-first writing removes that uncertainty.
The second part is scale. FIT Institute did not optimise one flagship page. It applied this structure across its entire course catalog — approximately 8 distinct lines. That matters because AI engines build entity confidence across multiple signals. Seeing consistent, answer-shaped content across a full catalog signals that this is a legitimate, authoritative source for this category.
2. Entity and topic authority across a complete catalog
AI engines need to identify who you are before they will cite you. FIT Institute's pages were consistent about the institution's name, the category of training it provides, and the industries it serves. This consistency appeared across the course pages, the about content, and the directory presence.
The multi-industry coverage — education qualifications, tax-agent certification, and fashion — reinforced topic authority rather than diluting it, because each area was supported by a coherent set of course pages rather than a single thin page. The AI could verify that FIT Institute genuinely covered each category in depth.
3. E-E-A-T signals that AI engines can read
Experience, expertise, authoritativeness, and trustworthiness are not concepts invented for AI — Google has used them for years. But AI engines read E-E-A-T differently from a human reviewer. They look for signals they can extract: specific claims with verifiable detail, named instructors or credentials, real course structures, and honest scope-setting about what a course does and does not cover.
FIT Institute's content included these signals at the course level. A page for a tax-agent program stated the accreditation context, the curriculum structure, and the candidate profile — not in vague marketing language, but in specific, extractable terms.
Generic content — "join thousands of successful graduates" — is close to unciteable, because the AI cannot verify it and cannot use it to answer a user's question. Specific content — course duration, credential outcome, target learner — can be extracted and attributed.
4. Passage-level citability
AI engines do not cite pages. They cite passages. This distinction changes how you structure content.
A passage is citable when it:
- Answers one question completely within a few sentences.
- Can be understood without surrounding context.
- Attributes its claims to a clear entity.
- Uses precise, not hedged, language.
FIT Institute's course pages were structured so that each section could stand alone as an answer. The section on "who should take this course" answered that question in two sentences. The section on "what certification does this lead to" named the certification directly. AI engines found clean extraction points throughout the content — and used them.
The technical upside: llms.txt and schema
FIT Institute earned its AI citations without two technical layers that are increasingly recommended for GEO: an llms.txt file and structured data schema markup.
An llms.txt file is a plain-text document at the root of your domain that tells AI crawlers which pages are most authoritative and how to interpret your content. Schema markup (particularly Course, Organization, and FAQ schema) gives AI engines machine-readable signals about your entity and your content structure.
Both are worth implementing. FIT Institute's result without them is evidence that content quality drives AI citation — and that these technical signals are multipliers, not prerequisites. If the content is not answer-shaped and entity-clear, llms.txt will not save it. If the content is already strong, adding these signals should improve both the volume and the consistency of citations.
What this means for your brand
The competitive logic of AI search is different from the competitive logic of traditional SEO. Domain authority and link volume still matter at the margins. What matters more is whether your content can be extracted, attributed, and used as the answer to a specific question.
A focused brand with well-structured content and clear entity signals can out-cite a larger competitor that is producing generic, brand-centric content — even in the same category, even for the same query.
The first step is an honest audit: for each service or product line, does your content answer the question a user would actually ask? Does the answer appear in the first two sentences of the relevant section? Can an AI engine identify who you are and what you offer from a single page, without reading your entire site?
If the answers are no, that is where GEO work starts.
FAQ
What is generative engine optimization (GEO)?
Generative engine optimization is the practice of structuring content and entity signals so that AI engines — including Google's AI Overview, ChatGPT, and Perplexity — cite your source when they generate an answer. It overlaps with SEO but rewards answer-first writing and entity clarity rather than keyword coverage alone.
Can a small or mid-size brand realistically out-cite a global firm like PwC?
Yes — as FIT Institute's result demonstrates. AI engines are not ranking brand size. They are selecting the most extractable, verifiable answer to a specific question. A well-structured page from a focused brand can be more citable than a generic page from a larger one, even in the same category.
Do I need llms.txt and schema to get cited by AI engines?
No — FIT Institute was cited without either. That said, both are worth implementing once your content is answer-shaped and your entity is clear. They add signal consistency and can improve the volume and reliability of citations.
How do I know if AI engines are currently citing me?
Test it manually: run the queries your customers would ask in ChatGPT, Gemini, and Google Search with AI Overview enabled. Note whether you are named, and if so, for which queries. That baseline tells you where you already have citation traction and where the gaps are.
Ready to audit your AI citation status?
If you want to know whether Google's AI Overview, ChatGPT, and Gemini currently mention your brand — and what it would take to change that — I offer a GEO audit that checks your entity clarity, your answer-first structure, and your passage-level citability across your key categories.
Book a call or send a message and we can look at your specific situation.