Case Studies Culinary & hospitality school
One program, one topic cluster: how a culinary school went from 205 to 3,660 Google AI Overview appearances in a quarter
A French culinary and hospitality school asked us for more search visibility on its programs. In five weeks we built 43 articles in four topic clusters, each one tied to a program it sells. Here is what moved, what didn’t, and what we’d tell a US vocational school considering the same play.
- ClientPrivate culinary & hospitality school, France (anonymized)
- Engagement3 months, SEO and conversion, started May 28, 2026
- MarketGoogle France, French-language search
- DataGoogle Search Console, read on October 20, 2026
Before and after, side by side
Same season, year over year: July 19 to October 17, 2025 against July 19 to October 17, 2026. Both periods include the back-to-school peak, so it does not inflate the result.
AI Overview appearances compare two consecutive quarters (April 20 to July 18, 2026, then July 19 to October 17, 2026), rather than two years. Google launched AI Overviews in France on July 22, 2026, so part of the increase reflects that launch. On the position bars, shorter is better.
Before
A school with real programs and almost no way to be found for them
Before the engagement, most of the school’s search traffic came from people who already knew its name. In the comparison season, searches that did not include the school’s name brought 392 clicks out of 2,200. A prospective student typing a program or a career question rarely landed on the school.
Three things stood out when we started:
- The flagship bachelor’s program sat on page 3 of Google, at an average position of 25.6. That is the program the school most wanted to fill.
- There was no editorial content at all. No articles, no guides, nothing answering the questions students ask before they pick a school. Zero pages for Google, or Google’s AI answers, to quote.
- Some programs had more than one address, which splits whatever authority a page builds between two URLs.
What we did
Five weeks of build, then the data had to catch up
-
May 28, 2026
Engagement starts
A three-month SEO and conversion engagement.
-
June 12
Strategy delivered
Audit, four program pages restructured, a content plan organized by topic, optimization for AI answer engines, and structured data.
-
June 16 to July 19
43 articles, 4 topic clusters
Each cluster maps to one program the school sells, with a hub page per topic linking articles to the program. Every program was consolidated onto a single URL, with 301 redirects from the old addresses.
-
Late August
First review
We delivered an August plan (a pastry cluster, a fix for the front-of-house program page). As of October 20, it has not been implemented. Keep that in mind when you read the next sections.
The detail
Where the growth actually came from
Same season, 2025 against 2026. We left brand searches (people typing the school’s name) out of the credit: that growth belongs to the school’s reputation, not to SEO.
Google AI Overviews
The pages we built are the pages Google’s AI quotes
AI Overviews are the generated answers Google now shows above the regular results. When a student asks a question about a career or a program, the school either appears in that answer or it doesn’t.
In the quarter after the build, the school appeared 3,660 times, up from 205 in the quarter before, with 69 different pages cited. More than half of those appearances come from pages that did not exist or were rebuilt during the engagement.
One piece of context matters: Google launched AI Overviews in France on July 22, 2026, three days into the second quarter. Part of the jump comes from that launch, which opened these answers to every site. What the work changed is which pages Google chose to quote.
The articles alone account for a third of every AI appearance the school gets, after five weeks of publishing.
Full transparency
What didn’t work
This is the part most case studies leave out. It is also the part that proves the mechanism.
Pastry: the only topic without articles, and the only topic that declined
The pastry articles were written but never published. Pastry is the one topic without a live cluster, and the one topic that went backwards: the pastry certificate’s clicks fell from 60 to 45 and impressions from 13,672 to 1,521, even though its page gained 16 positions. Same school, same period. The only difference is the missing cluster.
Front-of-house: the articles took the clicks from the program page
The topic grows as a whole, but the articles now win clicks the program page used to get: 25 down to 8 on the page itself. That is cannibalization, and it has not been fixed yet. Content that answers a question well can outrank the page that sells the program unless the two are deliberately separated.
An opportunity still on the table
Eight job-related searches (the equivalent of “server training” or “how to become a bar back”) generated about 6,800 impressions in three months for a single click. They rank on pages 2 and 3. That is the next quarter’s work.
What we are not claiming. This case study measures visibility and traffic. The school’s inquiries and enrollments were not connected to this data during the engagement, so we report no lead, enrollment or revenue figure here. Connecting search data to the school’s CRM is the next step we proposed.
These results come from Google France and French-language searches. We do not translate them into a US forecast.
For a US vocational school
What carries over, and what doesn’t
The market is different. The mechanism is the same one Google runs in the US: it ranks and quotes the pages that answer a student’s question, and it rewards a program page that sits at the center of related content.
Carries over
- One topic cluster per program you sell (HVAC, welding, cosmetology, medical assisting), each one linking back to its program page.
- One URL per program, with old addresses redirected.
- Articles written to be quoted: a direct answer first, then the detail. That is what AI Overviews pick up.
- A clear line between the article that informs and the page that enrolls, so they don’t compete.
Does not carry over
- The numbers. Search volume, competition and AI Overview coverage differ by country, state and trade. Your figures have to be measured on your market.
- The timeline. Five weeks of build and one quarter of data is what happened here, not a promise.
Your programs, measured
Which of your programs has no cluster behind it?
In a free session we look at your program pages, where they rank, whether Google’s AI answers mention you, and which program is the pastry case on your site: the one losing ground because nothing supports it.
Book Your Free Enrollment AuditBuilt exclusively for schools, colleges and training organizations.
