I screenshotted it before I believed it. FIT Institute, sitting in Google’s AI Overview as a cited source on a course query, with PwC further down the citation list. A specialist education brand out-cited a global firm on a term I’d been tracking. My first instinct was to celebrate. My second, the one that matters after thirteen years of reading dashboards that flatter you, was to ask what actually happened and how much of it would survive a retest.
So read this as the honest version, not the trophy version. FIT earned citations in Google’s AI Overview and ranked first for tracked course queries, including specific terms where it out-cited PwC. The lesson isn’t that a smaller brand can always beat a bigger one. It’s that a focused page becomes the preferred source when it answers the exact query clearly, shows topical fit, and hands the engine a verifiable passage to lift. Authority gets you considered. A clean answer gets you cited.
Here’s the take most agencies won’t print: chasing famous competitors is usually a vanity exercise. Out-citing PwC made a good headline, but the win came from owning one narrow question better than anyone else bothered to. This is a case-led teardown. It does not claim that schema, llms.txt, or one isolated change caused the result.
What the result proves—and what it does not
To put a number on it rather than a screenshot: across the 18 course queries I was tracking, FIT surfaced as a cited source in the AI Overview on roughly 13 of them, out-citing PwC Academy Middle East on specific terms, and the same pages were cited inside ChatGPT. That is a citation rate worth reporting, not a permanent score.
The captured evidence supports:
- FIT appearing as a cited source in Google’s AI Overview on about 13 of 18 tracked queries;
- first-place rankings for tracked course queries;
- out-citing PwC on specific terms;
- the result being observed in a defined search context.
It does not support:
- a permanent ranking or citation;
- the same answer for every location or user;
- a universal recipe that guarantees citation;
- attribution of the outcome to one technical feature.
That distinction matters. GEO reporting should preserve the query, date, locale, source URL, and screenshot rather than reduce everything to “AI visibility achieved.”
Why big brands win
Zoom out from FIT and PwC Academy Middle East for a moment, because the pattern behind this result generalises. On broad or ambiguous queries, AI Overviews default to the safe pick: the domain with existing authority, wide topical coverage, and enough public signal that the model already has an opinion about who it is. That’s not favouritism. It’s risk management on the model’s part — citing an established source is a defensible choice even when the passage itself isn’t the best one available.
Big brands also win on footprint alone. A firm the size of PwC Academy Middle East's parent network has hundreds of pages touching adjacent subjects, years of backlinks, consistent entity signals across the web, and a name other sources have already cited. On a wide query — professional certification courses in general, rather than one specific course code — that footprint tends to out-rank and out-cite a specialist, and there’s no trick that reliably beats it. If a plan depends on outranking a global brand on its broadest terms, budget for losing that particular fight. The opportunity for a smaller brand isn’t there. It’s one level down, on the specific questions the big brand’s broad pages were never written to answer.
Why the smaller source could be selected
The page matched the narrow question
Big brands carry broad authority and broad pages. Ask one of them a precise question and you often land on something written for everyone, which is to say written for no one in particular. The specialist wins the passage by being specific: a page that addresses the actual course, the actual requirement, the actual decision the searcher is weighing. PwC is not bad at this. It just wasn’t aiming at that question. FIT was.
The answer was easy to extract
The engine is lazy in a good way. It wants a passage it can quote without doing extra work, so the page that reads cleanly tends to win. Clear headings. Direct sentences. Definitions that stand on their own. A self-contained section beats a page that forces the model to gather scattered fragments and guess at the through-line. If a human skimming on a phone would have to scroll back up to understand the answer, the AI Overview will skip you too.
The site supported the entity
Related course pages, organisation information, internal links, and consistent naming help establish what FIT is and which topics it legitimately covers. This is ordinary entity and topical clarity, not a hidden GEO trick.
The content aligned with conventional SEO
Crawlability, indexability, query relevance, internal linking, and useful page content remain foundational. GEO did not replace SEO; the citation surfaced from work that also supported search visibility.
How small brands compete
The lesson from FIT’s citations isn’t “be smaller.” It’s “pick the fights a specialist can actually win.” A small brand doesn’t beat a global firm on breadth — it beats it on precision, freshness, and specificity, one query at a time.
Three positions where the odds favour the smaller brand:
- Narrow, high-intent queries the big brand’s page was never written for — a specific course code, a specific regulatory requirement, a specific comparison the searcher is actually making.
- Fresh or fast-moving facts a large, slower-moving site updates rarely — dates, fee changes, eligibility rules, local variations.
- Local or vertical specificity a horizontal brand won’t localise — country-specific requirements, sector-specific applications, language variants.
None of that requires beating PwC Academy Middle East, or anyone else, at their own game. It requires choosing questions where you’re the more credible answer and building a page that proves it. How to get cited in AI answers walks through the mechanics in more depth; the AI SEO and GEO page covers where this sits inside a broader search strategy.
The repeatable method
1. Build a query-to-page map
Assign each high-value course or buyer question to one canonical page. Consolidate pages that compete for the same intent.
2. Put the answer before the promotion
Open with the factual answer the searcher needs. Add the commercial proposition after the page has earned attention.
3. Add verifiable detail
Include requirements, steps, outcomes, dates, responsible entities, and limitations where relevant. Avoid unsupported superlatives.
4. Create supporting depth
Connect the answer page to related services, course details, FAQs, organisation information, and proof. Each supporting page should have its own job.
5. Remove technical ambiguity
Keep the page accessible in HTML, use stable URLs, correct canonical signals, sensible internal links, and structured data that matches visible content.
6. Capture the result properly
Record the query, market, date, surface, cited URL, and screenshot. Retest over time and report changes.
Citation assets
Not every well-written page earns a citation. The ones that do tend to share a shape:
- A definition or direct-answer block near the top — one self-contained paragraph that would still make sense if a model quoted only that.
- A comparison or requirement table — structured facts are easier for a model to lift cleanly than the same information buried in a paragraph.
- A dated fact sheet — fees, eligibility, deadlines, version numbers — anything with a date attached signals freshness and gives the model a reason to prefer the page over a stale competitor page.
- An FAQ-formatted section that mirrors how people actually phrase the question, not how the organisation prefers to phrase it.
Build these as the load-bearing part of the page, not an appendix bolted on afterward. One strong citation asset, on the right query, tends to outperform a whole content calendar of general posts.
Measurement
Citations move. The FIT result held on roughly 13 of the 18 tracked queries at the time of capture — not all 18, and not permanently. Before claiming GEO progress anywhere, build the measurement habit before the page.
At minimum, log the exact query, the market and language, the date, the surface (Google AI Overview, ChatGPT, Perplexity, Bing Copilot), the cited URL, and a screenshot. Retest on a schedule rather than once after launch — citations that appear in week one can disappear by week four, and the reverse also happens. Measuring AI search visibility covers the tracking setup in more depth, including what to log when a citation moves.
A single screenshot is not proof of a durable win. A rate, over a defined set of queries, on a defined date, retested and updated, is closer to honest reporting.
What not to copy from this case
Plenty of teams will read a result like this and reach for the wrong moves. Don’t spin up near-duplicate pages for every phrasing of the same question; you’ll cannibalize yourself and teach the engine you have nothing distinct to say. Don’t take a competitor’s answer and paste your brand over it. The engine has already seen the original, and the imitation rarely earns the citation. Don’t bolt on schema that claims facts your visible page doesn’t support, and don’t mistake one citation for a permanent win. I’ve watched citations appear and quietly disappear on retest; that’s the surface working as designed, not a betrayal.
The one I’d underline: stop hunting famous competitors and start answering buyer-relevant questions. Out-citing a big name is a side effect of doing the boring work well, not a strategy you can aim at. And don’t assume FIT’s page structure will win you the same query without the same underlying evidence behind it. The transferable advantage is editorial precision, not mimicry.
A buyer’s application
If you lead an education or professional-services business, choose one commercially important question where:
- the current result is broad or incomplete;
- your organisation has direct expertise;
- you can provide a more specific answer;
- the answer naturally connects to a service or programme;
- you can monitor it without exaggerating the result.
Build that page first. One well-owned query is a better test than a site-wide “GEO rewrite.”
How this case connects to the wider system
This post owns the specific FIT-versus-PwC observation. The full evidence belongs in the FIT GEO case study. The general implementation checklist is in how to get cited in AI answers. The service-level scope is in AI SEO and GEO.
Next step
If you want to identify the questions where your specific expertise can outrank or out-cite a broader brand, request a systems diagnostic.
Find your AI citation gaps. A short diagnostic run against your own tracked queries shows which ones are realistically winnable and which ones aren’t worth chasing — start here.