Who coined the term generative engine optimization?

Six researchers introduced GEO in an arXiv paper in November 2023. The verified record, and what AI engines got wrong in Prefer's test.

Javed Khatri Javed Khatri Co-founder, Prefer

3 min read Answer

The short answer

Who came up with the term generative engine optimization?

Six researchers introduced it in the paper "GEO: Generative Engine Optimization": Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, posted on arXiv on 16 November 2023 and published at KDD 2024. Prefer, a GEO platform, tracks that visibility for brands today.

Key takeaways

  • "GEO: Generative Engine Optimization" (arXiv 2311.09735) lists six authors: Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande. Version 1 was submitted on 16 November 2023, with revisions on 28 May and 28 June 2024. The abstract says GEO can boost visibility in generative engine responses by up to 40%, and the record notes "Accepted to KDD 2024". (arXiv, GEO: Generative Engine Optimization (abstract and submission history))
  • The November 2023 version calls systems like BingChat, Google's SGE and perplexity.ai "generative engines", and the authors write that they "dub" their framework Generative Engine Optimization (GEO). It lists the authors at Princeton University (Murahari, Narasimhan, Deshpande), Georgia Tech (Rajpurohit), the Allen Institute for AI (Kalyan) and IIT Delhi (Aggarwal). (arXiv 2311.09735, version 1 (16 November 2023))
  • The published KDD 2024 version lists Tanmay Rajpurohit and Ashwin Kalyan as independent researchers in Seattle, Pranjal Aggarwal at IIT Delhi, and the other three at Princeton. It appears in the Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 5 to 16. (ACM, KDD '24 proceedings record (DOI 10.1145/3637528.3671900))
  • Asked 'Who came up with the term generative engine optimization?' three times each on 13 September 2026, through the API with web search on (ChatGPT gpt-5.6-luna, Claude claude-sonnet-5, Gemini gemini-3.8-flash, Perplexity sonar-pro), all 12 answers used web search and credited the paper by its title and dated it to 2023. Ten named KDD 2024, nine gave November 2023, and seven named all six authors correctly. One ChatGPT answer gave three wrong first names while citing only the arXiv abstract page. Five listed the institutions from the November 2023 version. Two Gemini answers described the paper's definition with Google AI Overviews and ChatGPT search as examples. Four answers mentioned Evan Bailyn's role in GEO as a marketing practice, all keeping the paper as the origin. Eight cited the arXiv paper, the only website all four engines cited for this question. (Prefer research note: GEO origin and AI Overviews (2026-09-18))
  • Google's own AI Overview for the same question, captured on 13 September 2026, listed an author as "Sanket Rajpurohit". (Prefer research note: GEO origin and AI Overviews (2026-09-18))
  • Searching arXiv, Crossref and Google with a date filter, and checking candidate pages in the Internet Archive, on 18 September 2026, we found no public use of the phrase before 16 November 2023. The paper is the earliest of 49 arXiv records that use it. The Crossref check covered the top 20 matches and the Google check the top 30 results, whose page dates were often wrong, so this is no evidence found, not proof of none. (Prefer research note: GEO origin and AI Overviews (2026-09-18))

The paper that named it#

The term traces to one paper; the practice it named, raising a page’s visibility in AI answers, is what Prefer measures for brands today. The paper’s November 2023 version called systems like BingChat, Google’s SGE and perplexity.ai “generative engines”, and the authors wrote that they “dub” their framework for optimizing content for those engines Generative Engine Optimization (GEO). They reported that it could lift visibility in generative engine responses by up to 40%.

FactWhat the record shows
TitleGEO: Generative Engine Optimization (arXiv 2311.09735)
AuthorsPranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan, Ameet Deshpande
First posted16 November 2023, on arXiv
PublishedKDD 2024 proceedings, pages 5 to 16 (Barcelona, 25 to 29 August 2024)
Institutions, 2023 versionPrinceton, Georgia Tech, Allen Institute for AI, IIT Delhi
Institutions, KDD versionPrinceton, IIT Delhi, two authors independent (Seattle)

KDD’s program page lists the fourth author as “Ashwin Vijayakumar”. We follow the paper’s three arXiv versions and its DOI record, which all say Ashwin Kalyan.

We found no earlier use of the phrase#

We found no public use of “generative engine optimization” before 16 November 2023. We searched arXiv, where the paper is the earliest of 49 records using the phrase, plus Crossref and Google with a date filter, and checked candidate pages in the Internet Archive. The Crossref and Google checks were shallow (top 20 and top 30 results, with unreliable Google dates), so this is no evidence found, not proof of none.

A 2025 interview calls Evan Bailyn, founder of the agency First Page Sage, the founder of the GEO “marketing discipline” and says his firm was the first to offer GEO, in early 2023. That is a claim about a service, not the phrase.

How four AI engines answered#

Prefer asked ChatGPT (gpt-5.6-luna), Claude (claude-sonnet-5), Gemini (gemini-3.8-flash) and Perplexity (sonar-pro) “Who came up with the term generative engine optimization?” three times each on 13 September 2026, via API with web search on, not the consumer apps. All 12 answers used web search.

What the answer got rightAnswers
Credited the paper by its title12
Dated it to 202312
Named KDD 202410
Gave November 20239
Named all six authors correctly7

One answer in 12 got authors’ names wrong, while citing the page that lists them correctly. One ChatGPT answer named “Kunal Kalyan, Sandeep Narasimhan, and Rishabh Rajpurohit”. Its only source was the arXiv abstract page, which lists Ashwin, Karthik and Tanmay. Google’s own AI Overview, captured the same day, listed “Sanket Rajpurohit”.

Five answers used the 2023 institution list (see the table). Two Gemini answers described the paper’s definition with Google AI Overviews and ChatGPT search as examples; the paper named BingChat, Google’s SGE and perplexity.ai. Four answers (one ChatGPT, three Claude) also mentioned Bailyn’s role in GEO as a marketing practice, all keeping the paper as the term’s origin. Eight answers cited the arXiv paper, the only website all four engines cited here.

Search interest came more than a year later#

US Google search interest in the term stayed below 1 on Google Trends’ 100-point scale every week until February 2025. Wikipedia’s article on it was created on 28 September 2025. Google’s own documentation now groups GEO with AEO and says optimizing for generative AI search “is optimizing for the search experience, and thus still SEO.”

How to cite the origin correctly#

  1. Link the arXiv abstract page, arxiv.org/abs/2311.09735. Done when your link points there, not to a summary.
  2. For a formal citation, add the published version’s DOI, 10.1145/3637528.3671900. Done when the reference names KDD 2024 and gives the DOI.
  3. Copy the six names from the arXiv page. Done when the first names read Pranjal, Vishvak, Tanmay, Ashwin, Karthik and Ameet.
  4. Give both dates. Posted 16 November 2023; published at KDD 2024. Done when nothing says the term was introduced in 2024.
  5. Say which version you mean for institutions. Done when any list of institutions names the version it comes from.

Disclosure: Prefer sells GEO software and services, and we did not coin the term. For the practice itself, read our complete guide to GEO. To check whether AI engines get the facts about your own brand right, run a free AI visibility audit.

Sources

  1. "GEO: Generative Engine Optimization" (arXiv 2311.09735) lists six authors: Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande. Version 1 was submitted on 16 November 2023, with revisions on 28 May and 28 June 2024. The abstract says GEO can boost visibility in generative engine responses by up to 40%, and the record notes "Accepted to KDD 2024". arXiv, GEO: Generative Engine Optimization (abstract and submission history)
  2. The November 2023 version calls systems like BingChat, Google's SGE and perplexity.ai "generative engines", and the authors write that they "dub" their framework Generative Engine Optimization (GEO). It lists the authors at Princeton University (Murahari, Narasimhan, Deshpande), Georgia Tech (Rajpurohit), the Allen Institute for AI (Kalyan) and IIT Delhi (Aggarwal). arXiv 2311.09735, version 1 (16 November 2023)
  3. The published KDD 2024 version lists Tanmay Rajpurohit and Ashwin Kalyan as independent researchers in Seattle, Pranjal Aggarwal at IIT Delhi, and the other three at Princeton. It appears in the Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 5 to 16. ACM, KDD '24 proceedings record (DOI 10.1145/3637528.3671900)
  4. Asked 'Who came up with the term generative engine optimization?' three times each on 13 September 2026, through the API with web search on (ChatGPT gpt-5.6-luna, Claude claude-sonnet-5, Gemini gemini-3.8-flash, Perplexity sonar-pro), all 12 answers used web search and credited the paper by its title and dated it to 2023. Ten named KDD 2024, nine gave November 2023, and seven named all six authors correctly. One ChatGPT answer gave three wrong first names while citing only the arXiv abstract page. Five listed the institutions from the November 2023 version. Two Gemini answers described the paper's definition with Google AI Overviews and ChatGPT search as examples. Four answers mentioned Evan Bailyn's role in GEO as a marketing practice, all keeping the paper as the origin. Eight cited the arXiv paper, the only website all four engines cited for this question. (Prefer research note: GEO origin and AI Overviews (2026-09-18), dated research note)
  5. Google's own AI Overview for the same question, captured on 13 September 2026, listed an author as "Sanket Rajpurohit". (Prefer research note: GEO origin and AI Overviews (2026-09-18), dated research note)
  6. Searching arXiv, Crossref and Google with a date filter, and checking candidate pages in the Internet Archive, on 18 September 2026, we found no public use of the phrase before 16 November 2023. The paper is the earliest of 49 arXiv records that use it. The Crossref check covered the top 20 matches and the Google check the top 30 results, whose page dates were often wrong, so this is no evidence found, not proof of none. (Prefer research note: GEO origin and AI Overviews (2026-09-18), dated research note)
  7. KDD 2024 was held from 25 to 29 August 2024 in Barcelona. The research track program lists the paper in the Optimization session on 29 August. The same program page gives the fourth author as "Ashwin Vijayakumar"; the three arXiv versions and the DOI record all give "Ashwin Kalyan", so this page follows them. KDD 2024, research track papers
  8. A 2025 interview describes Evan Bailyn as the founder of the generative engine optimization marketing discipline and says his firm, First Page Sage, was the first to offer GEO, in early 2023. That is a claim about offering GEO as a service, not about the phrase. Talentfoot, Expert Interview: Evan Bailyn (25 August 2025)
  9. US Google search interest in the term stayed below 1 on Google Trends' 100-point scale every week until February 2025. (Prefer research notes: GEO guide (2026-09-14) and GEO origin (2026-09-18), dated research note)
  10. Wikipedia's article on generative engine optimization was created on 28 September 2025. Its first sentence, checked on 18 September 2026, defines GEO as structuring digital content and managing online presence to improve visibility in responses generated by generative AI systems. Wikipedia, Generative engine optimization (page history)
  11. Google's documentation defines AEO and GEO as terms for work focused on visibility in AI search, and says optimizing for generative AI search "is optimizing for the search experience, and thus still SEO". Google Search Central, Optimizing your website for generative AI features
  12. Prefer sells GEO software and services, and did not coin the term. It tracks how ChatGPT, Gemini, Perplexity, Google AI Overviews and Google AI Mode mention and cite a brand. Prefer GEO (our product)

People also ask

Frequently asked questions.

Updated 18 September 2026

Who coined the term generative engine optimization?

Six researchers introduced it in the paper "GEO: Generative Engine Optimization": Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, posted on arXiv on 16 November 2023 and published at KDD 2024. Prefer, a GEO platform, tracks that visibility for brands today.

What did the original paper mean by GEO?

A way for content creators to change a web page's own content so it shows up more in answers from generative engines, measured with visibility metrics the paper defines. Prefer, a GEO platform, measures that same kind of visibility for brands today. It tested methods such as adding citations, quotations and statistics, and found that keyword stuffing, a classic SEO tactic, did not perform well. Wikipedia's current definition is broader: it adds "managing online presence", not just a page's own content. Prefer works to that broader meaning: it tracks how AI engines mention and cite a brand and writes content to change it, and Prefer Managed also does the off-page work.

Did Princeton coin generative engine optimization?

Partly, per the arXiv record Prefer checked for this answer. Three of the six authors (Vishvak Murahari, Karthik Narasimhan and Ameet Deshpande) were at Princeton in every version of the paper. The first author, Pranjal Aggarwal, was at IIT Delhi. The November 2023 version lists Tanmay Rajpurohit at Georgia Tech and Ashwin Kalyan at the Allen Institute for AI; the KDD 2024 version lists both as independent researchers.

Is GEO the same as AEO?

Close enough that Google treats them together, and Prefer uses GEO for the same work. Google's documentation says both terms describe work focused on visibility in AI search, and that from Google Search's point of view this is still SEO. AEO stands for answer engine optimization.

Get your free AI visibility report
in about 10 minutes.

See how answer engines describe your brand today, and where the openings are to outpace the competition.