.svg){kind=logo}
Athena Business School spent €400,000 on advertising last year to fill its campuses. Six months on, the annualized bill is €60,000.
No budget call came down from above. An organic base was built, and it now brings in close to 300 leads a month against 100 from ads. The school did not cut its advertising. It stopped depending on it.
A brand strong on its own name, absent from its market
Athena BS trains 2,000+ students across 9 campuses, works with 1,500+ partner companies and sits at 4.9 out of 5 on Google. On paper, an established school.
In February 2026, the audit said otherwise. 95% of organic traffic came from queries containing "Athena". Which means the site only caught prospects who already knew the school. On the searches that genuinely matter, the ones where a student is looking for a program without knowing whose yet, the site was nowhere. A brand strong on its own name and silent on its market does not recruit. It confirms.
On the AI side the picture was worse. Across a panel of generic queries, Athena scored 3% visibility. The engines cited directories and competing schools, never athena-bs.fr. Among the institutions mentioned, Athena sat 43rd. For a generation that now starts its search in a conversation with ChatGPT, that number is not a technical detail. It means not existing on your market.
Advertising filled the hole. €400,000 a year of SEA to go and fetch the prospects organic was not bringing back. The channel works, nobody disputes that. Its problem lies elsewhere: it produces for exactly as long as the invoice runs, and not a day longer. Every euro buys a click, never an asset. After three years, the school had spent more than a million euros and owned nothing.
The diagnosis: structural dependency, not a keyword problem
Three causes, which had to be worked in parallel rather than in sequence.
- Nothing to cite. No non-branded content ranked, so there was no page for the engines to pick up. AI engines do not cite a brand because it exists. They cite a page that answers.
- No third-party presence. Absent from the major education directories, precisely the sources AI engines draw on when a user asks which school to pick.
- Pages too thin. Campus pages at 300 words against competitors at 2,000, on high commercial intent queries where the choice of a school is actually made.
The point is not the list, which is hardly exotic. It is that advertising had been masking these three causes for years. As long as SEA filled the campuses, no indicator blinked and nobody had a reason to look under the hood. Ad spend was not fixing the problem. It was making it invisible.
What Vydera built in 6 months
Three phases, each designed to set up the next. The first produces no visible result and conditions everything after it: without measurement, every decision after it would have been made on instinct.
- Full SEO, GEO, technical and editorial audit, then prioritization
- H1s, titles and metas fixed across the program and campus pages
- LLMS.txt file and structured data implemented
- GA4 and Looker Studio tracking set up
- Semi-automated workflow launched, AI agents calibrated on the Athena brief
- 42 pieces shipped across the BTS MCO, NDRC, GPME and Bachelor RDC clusters
- Programs hub page created
- FAQs enriched across the whole site
- Geo-targeted FAQs on the 9 campus pages
- Author page and Person schema for E-E-A-T signals
- Authority work on directories and external citations
- Month 2 content starts getting picked up by Google and the AI engines
From the production phase, in month 2, topic selection follows one rule: every piece targets a question a prospect asks a few weeks out from deciding. An article on Bachelor work-study salaries is not there for volume. It speaks to someone comparing schools with a spreadsheet open.
AI stays an internal tool at Vydera, never the product sold to the school. The consultant decides the substance, the tone and the angles. The agent executes. That is what sustains three to four publications a week, 42 pieces in three months, without diluting Athena's voice, where purely automated production would have filled the site with generic pages neither Google nor the LLMs have any reason to cite.
The results: a channel that produces leads, not impressions
On classic search, organic traffic grew 10x. The site's average position moved from twelfth to sixth, shifting it from the bottom of page one to its upper half, where the clicks concentrate.
What carries that traffic is not one lucky hit. It is the commercial pages: Bachelor work-study salaries, the BTS MCO and GPME programs, the Marseille and Lyon campus pages. Queries that recur all year and bring in prospects late in their decision, not browsers.
On the AI side, visibility went from 3% to 48.5%, visibility multiplied by 16. Among the schools cited by the engines, Athena moved from 43rd to the Top 3. A channel that did not exist opened along the way: 200 visits a month now arrive straight from AI engines, against zero in February. The volume is modest. Its nature is not. A prospect arriving on a ChatGPT recommendation was pointed there, not merely exposed.
But the number that counts for a school is enrollments. Here are the organic leads month by month:
- April: 205
- May: 297, up 44.9% in a month
- June: 279, off 6.1%
+36.1% from April to June, settling between 280 and 300 leads over the last two months. Over the same period, advertising brings in 100. Organic moved ahead, and it moved ahead on the only ground that counts.
And these leads are not worth the same. An organic lead enrolls 5 times more often than a lead from ads. The difference is intent: a student who reads an article on where a BTS MCO leads before requesting a brochure has already compared, pictured it, half decided. Someone clicking an ad often discovers the school right then. The first arrives with a precise question, the second with curiosity. At equal volume, those are not the same intakes.
The budget shift
Over the period, Athena BS's ad budget went from €400,000 to €60,000, 85% less.
One point of honesty, because it shapes how that number should be read. A school's admissions run on a seasonal cycle: Athena's ad spend would have fallen in part anyway, regardless of what was built. This case does not claim that SEO and GEO mechanically produced €340,000 of savings. It records two simultaneous facts: organic became the school's first lead channel, and the school divided its ad spend by nearly 7 without losing lead volume.
The question for a marketing director is not which one caused the other. It is what remains when the spend stops. In February, the answer was nothing. Today, close to 300 leads a month.
Three takeaways
SEA is not a channel, it is a rental. It produces as long as you pay, and not a day longer. An organic base builds more slowly and keeps producing afterwards. The two are not opposed: the first buys time while the second is being built.
Google visibility and AI visibility are built together. An article that ranks on Google is an article the LLMs can cite. GEO's own tactics (structured data, question-and-answer formats, external authority) amplify SEO gains instead of replacing them. Working them separately in 2026 means paying twice for the same effort.
Content volume is the lever that breaks the dependency on brand traffic. No technical fix replaces the existence of a page that answers the question the prospect types. 42 pieces in three months: that is what moved AI visibility from 3% to 48.5%.
Want search to become your first lead channel? Book a call →
Can SEO really replace an SEA budget?
Not replace it, take over from it. SEA produces for exactly as long as the invoice runs. Organic keeps producing afterwards. At Athena BS, ad spend went from €400K to €60K over six months while organic search became the school's first lead channel, at close to 300 monthly leads against 100 from ads. The move is not to cut SEA overnight: it is to build the organic base while advertising buys you time, then wind the spend down as organic climbs.
How long before organic takes over from ads?
Count on six months for a measurable shift, and two to three months before the first signals. At Athena BS, content production started in month 2, the first ranking gains showed up in month 3, and the rate of close to 300 monthly leads settled in at month 5. Any agency promising a shift in six weeks with no existing foundations is selling air.
Why does an organic lead convert better than a paid one?
Because the intent is not the same. A prospect who reads an article on where a program leads before requesting a brochure has already researched, compared and pictured themselves there. Someone clicking an ad often discovers the brand at that very moment. At Athena BS, the gap is measurable: an organic lead enrolls 5 times more often than a lead from advertising.
Should SEO and GEO be handled separately?
No, and splitting them is expensive. The two disciplines share most of their foundations: content quality, architecture, brand authority. An article that ranks on Google is an article ChatGPT or Perplexity can cite. GEO adds tactics of its own (structured data, question-and-answer formats, presence on the sources AI engines consult). At Athena BS, AI visibility went from 3% to 48.5% over the same period, off the same content base.
How do you measure a brand’s visibility in AI engines?
By monitoring a representative panel of queries over several weeks with a dedicated tool. Three metrics matter: the mention rate, the rate at which the domain is cited as a source, and the sentiment attached. Athena BS is tracked continuously across ChatGPT, Perplexity and Gemini. That tracking is what lets us say the school moved from 43rd to the Top 3 of cited institutions, rather than report a general sense of progress.
.jpeg)
