Three recent studies (Aggarwal/Princeton, BrightEdge, seoClarity) and the voices of Solis, Ray and Illyes document a measured reading of the GEO vs SEO debate.
The debate on optimization for generative AI engines is now taking place in the boardroom. A recent column published in these columns defends the idea of a new AI visibility market, which must be conquered before it crystallizes. The thesis is attractive. However, it poses an observation that deserves to be compared: that of a break between the SEO of yesterday and a new discipline, called GEO (Generative Engine Optimization). Three recent studies tell a different story, and the voices which really structure the international debate converge towards a reading opposite to that which agitates certain French councils.
GEO is not a replacement for SEO. This is its extension in a new display context.
What the founding paper really says
The study that popularized the term GEO, signed by Pranjal Aggarwal (IIT Delhi) and co-authors from Princeton at KDD 2024measures more than 40% visibility gain in generative engines thanks to techniques called Generative Engine Optimization. When you open the paper, the methodology is self-explanatory: the authors built GEO-bench, a test bench of 10,000 queries covering various domains and sources. They tested nine distinct optimization strategies, compared on several generative models, and measured the effect of each on the visibility of the source in the written response.
The winning levers can be summed up in a few lines: insertion of verifiable and attributed quotes, numerical statistics, authority signals, factual formulation, etc. No experienced specialist will read this list as a deal breaker. This is mature SEO, transposed to a context where the response is written by a model rather than displayed in a blue list. Nowhere does the paper say that SEO is dead. It describes a continuum optimization for a new restitution format.
Three competing vocabularies are in circulation to designate these levers: GEO, AEO (Answer Engine Optimization) and AISO (AI Search Optimization). The operational scope remains the same.
The measured overlap, three converging studies
This is where the numbers make the difference. BrightEdge measured over sixteen months, from May 2024 to September 2025, across nine industriesthe share of AI Overviews (AIO) citations coming from Google organic ranking: 54.5% of URLs cited by AIOs are in the organic top 100, compared to 32.3% at the start of the measurement. The curve goes up, not the other way around. To qualify: only 16.7% of citations are strictly in the top 10. The top 100 captures the majority, the strict top 10 remains in the minority.
The largest study to date comes from seoClarity, on 362,000 American keywords analyzed in October 2025 : 94% of AIOs cite at least one URL present in Google’s organic top 20, and the overlap with the top 10 reaches 32% of citations. If the AIO draws 94% in the Google top 20, talking about infrastructure to be rebuilt from scratch contradicts the figures.
There remains the lost click objection. Ahrefs, on 300,000 keywords and with an update of the initial study in December 2025measures an average drop of 58% in CTR in position 1 when an AI Overviews appears, continuously deteriorating since the 34.5% measured in the spring. The number is scary if read alone. Taken with the two previous ones, it says something else: the click does not disappear completely, it is partially redeployed towards the sources cited in the AIO. Publishers who appear in the panel get a residual share of traffic, those who do not appear there exit the game.
The voices that structure the debate
Aleyda Solis, international SEO consultant, founder of Orainti and keynote speaker at BrightonSEO, SMX and MozCon for almost fifteen years, in her report from Search Central Live Zurich 2025poses the phrase that has been circulating since: “There is no GEO without the implementation of SEO fundamentals.” She extends the same reading in an interview for Advanced Web Ranking where she describes the work as applying existing SEO rules to a new platform, not replacing them.
Lily Ray, VP SEO & AI Search at Amsive and reference voice on EEAT since the Helpful Content Updateextends the analysis in a reflection published in January 2026: “AEO and GEO are neither an overhaul nor an abandonment of SEO. On the contrary, they represent a new system for competing, capturing and measuring success on AI platforms.” Ray specifically targets what she calls “GEO grifters”, these consultants who repackage classic SEO fundamentals under a new name and resell them at a high price.
Jeremy Moser, co-founder and CEO of the uSERP agency, interviewed by Digiday on the GEO bubbledecides even harsher: “If a GEO service doesn’t tell you frankly that success in AI visibility is 80% good fundamental SEO, they are selling you snake oil.” The formula has become the marker of practitioners who refuse the narrative of rupture.
The clearest signal comes from Google. During the July 2025 Search Central Live Deep Dive Asia Pacific, Gary Illyes, Search Relations analyst at Google, relayed by Search Engine Journalconfirmed that normal SEO practices are enough to make content appear in AI Overviews. No parallel discipline to add.
The publishing profession has already seen these “ruptures” pass
When we have been operating online media for thirteen years, we have seen a few cycles go by. Clickability first for the golden age of Facebook, mobile first and the Google AMP or Facebook Instant Articles period, video first and the boom of native media: with each new restitution format, consultants declared a break, sold an infrastructure to be rebuilt, and came back two or three years later to explain that the fundamentals had only been extended to a new context. The budgets committed to the “disruption” have rarely produced the promised return. Those who had maintained investment in the fundamentals of the profession (the right content, at the right time, in the right format, distributed to the right target, on as many media as possible) capitalized on each extension.
The pattern repeats itself every cycle: oversell the break to open a separate advice line, and in the process, underinvest in the bricks that continue to pay. The announced breakup always presents itself with a new acronym, its own vocabulary, and a sense of urgency that makes it impolite to ask for the measure. Editors who have the history of the profession know how to recognize the sequence.
The right question, for a leader looking at the bill, is not “how do I align with these new narratives?”. It’s “what do the numbers say today, and what continues to pay?”. On the GEO subject, the measured response is documented in the three studies cited above.
A context that is evolving at high speed
The display context is evolving at high speed, and the original forum is correct on several points. BrightEdge measures AIO penetration at 48% of queries tracked, up 58% year-on-year. Bots from generative platforms (GPTBot, ClaudeBot, PerplexityBot) must be explicitly authorized in robots.txt otherwise their responses will be invisible. Organization or LocalBusiness Schema.org markup gains value. The multi-surface presence matters more than at the time of just ten blue links. Reputation as perceived by models, their tendency to confuse, hallucinate or report in a biased manner, becomes a subject of surveillance in its own right. This involves adding two requirements to the existing SEO profession, treating AI agents as crawlers in their own right and measuring the mature click after displaying an AIO, not recreating what would be a “new discipline”.
The worst service to render to a manager who is discovering these subjects is to describe to him a virgin market where his competitors would not yet be. Google positions are already taken by well-ranked sites, and AIs copy them by more than 50%. The work to be done is measurable, executable, marginally new but very largely continuous. No reason to sign a separate GEO budget line. SEO will not be replaced by GEO, since GEO… remains SEO.
