French entrepreneur Romain Brunel says his sofa comparison site’s AI referral traffic rose from single digits to an average of 267 visits per day in one month through a GEO strategy
On November 9, 2026, French entrepreneur Romain Brunel (@Affiseo_) posted on X that after applying a generative engine optimization (GEO) strategy to his sofa comparison website Quel Canapé starting on October 11, the average daily traffic from AI rose from about 5–20 visits to 267 visits on November 8 (with ChatGPT contributing the vast majority), and the growth curve is still rising. He said these visitors are all highly qualified traffic.
Brunel’s specific measures included: first, making sure the website is indexed by Bing, because he said ChatGPT defaults to Bing when it needs to search the web; second, optimizing pages for ultra-long-tail, highly specific queries (for example, “Matelas 160x200 Ressorts Ensachés,” “Canapé Couchage Quotidien,” “Canapé Best Mobilier Panoramique,” and so on) in order to precisely match users’ search intent in ChatGPT. He gave an example saying that one page ranked for “Meilleur canapé Best Mobilier Panoramique”; Semrush showed the search volume was close to 0, but it still brought in 4 qualified visits in a single day, and such queries together contributed to the day’s 267 visits. He further said that such pages also have relatively high exposure opportunities on Google because of less competition, and that the strategy is applicable to affiliate marketing and e-commerce sites.
Brunel is the founder of Quel Canapé (quel-canape.fr), an independent comparison platform for sofas, armchairs, and mattresses, operated by AFFISEO and mainly monetized through affiliate commissions. A screenshot of the analysis attached to the post showed 267 AI visits on November 8, while many days in mid-October were in the single digits to more than 20, which is consistent with his statement. The above traffic and conversion data were all self-reported by him, and there has been no third-party independent audit seen so far.
Original source:
Title: C'est facile d'obtenir du trafic (beaucoup de trafic) depusi Chat GPT... Publisher: Romain Brunel - Affiseo (@Affiseo_) Publication date: 2026-11-09 URL: https://x.com/Affiseo_/status/2119679286193852688 Nature of source: first-party original post (including analysis screenshots) Verification status: The full original post and attached screenshots were read; the traffic figures are self-reported, and the site’s existence and founder identity were checked. Supporting content: GEO strategy description, traffic growth figures, Bing and long-tail query recommendations, Quel Canapé example.
ABAB AI Insight
Brunel’s case moves GEO from conceptual discussion to an observable level of daily traffic, but its core logic depends on ChatGPT’s continued reliance on Bing indexing in its retrieval path. Public information in 2026 shows that ChatGPT search has shifted to OpenAI’s own index (OAI-SearchBot) and a hybrid model with third-party providers; Bing remains an important source, but not the only one or an always-default channel. Some research also indicates that the share of Google-crawled results in external results is rising. Therefore, the statement that “simply getting indexed by Bing is enough to obtain ChatGPT traffic” needs to be limited to “helpful for entering the retrieval candidate pool,” rather than being treated as a sufficient condition.
The effectiveness of the ultra-long-tail strategy is logically consistent for low-competition queries: when users pose highly specific product model or specification questions in a conversation, pages with high match quality are more likely to be cited and drive clicks. This partly overlaps with the traditional SEO experience that long-tail keywords convert better, but citation decisions in AI answers are also affected by page structure, authority signals, and retrieval snippet quality; keyword coverage alone is not enough to guarantee results. As a comparison platform, Quel Canapé already has structured comparison and specific specification information in its content, which may have amplified the effect of long-tail matching.
From a business perspective, the growth occurred at the same time as Brunel publicly mentioned site restructuring and multichannel expansion (YouTube, Pinterest, etc.), so it is difficult to fully separate the contribution of other factors. He is also promoting AI automation tools and courses such as Affiseo Hub, so the post has a case-marketing angle. For affiliate or e-commerce sites, long-tail pages can reduce competition costs for head keywords, but scaling depends on content production efficiency; if relying on AI-generated content, quality and update maintenance costs must be watched to avoid ranking downgrades by retrieval systems or a decline in user experience.
Overall, this self-reported data suggests that AI referral traffic is becoming a trackable independent channel, and that in certain verticals targeted content can secure early gains, but its replicability and sustainability are still constrained by changes in AI retrieval architecture, intensifying competition, and the site’s own conversion capabilities. There is still insufficient evidence to directly generalize the short-term growth of a single site into a universal conclusion that “GEO can completely ignite a business.”