Axis

Three measures

Updated

SEO, GEO, and AEO.

SEO, GEO and AEO describe one search space, read with three different measures. SEO (search engine optimization) asks whether a page is found, ranked and chosen from a list of results. Google describes it as helping search engines understand your content and helping people decide whether to visit [1]. GEO (generative engine optimization) asks whether a page is cited, and how much of it shows, inside an answer that a language model writes. The term comes from a 2024 paper with a benchmark of 10,000 queries [2]. AEO (answer engine optimization) is a label used in the industry for being the source of a direct answer. In the sources we checked, we found no founding paper and no shared definition for it. So we treat AEO as a marketing term, not a field of study.

What each term measures

Three names for one search space
TermWhat is measuredWhere the definition comes fromStatus
SEO Position and click in a list of links Search engine documentation, for example Google Search Central [1] Established practice, documented by the engines themselves
GEO How much of a generated answer is attributed to a source, weighted by position, plus a judged impression score [2] Aggarwal et al., KDD 2024 [2] Research term with a public benchmark. The literature is young.
AEO Being the source a direct answer is drawn from Industry usage Marketing label. No canonical definition found.

SEO: the list still sits underneath

Google’s own starter guide is plain about limits: “There are no secrets here that’ll automatically rank your site first in Google” [1]. Its page on AI features goes further. It says the best practices for SEO remain relevant for AI Overviews and AI Mode, and that there are no additional requirements to appear in them [3]. To be shown as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet [3]. The same page says you don’t need new machine-readable files, AI text files or special schema.org markup to appear [3]. For Google’s AI features, then, the ordinary SEO surface is the entry ticket to the generated one.

GEO: a measure built for answers

Aggarwal and colleagues describe a generative engine as a system that retrieves documents and uses a large model to write a response grounded in those sources, with inline attributions [2]. Visibility there is not a rank. They measured it with a position-adjusted word count and a subjective impression score [2]. On their benchmark, adding citations, quotations and statistics raised the word-count measure by about 30 to 40 percent. Adding more of the query’s keywords offered little or no improvement [2]. That is the same finding our note on generated answers reports.

The recipes are less settled than the measure. A 2025 benchmark, C-SEO Bench, tested published editing methods across two tasks and six domains, and with several competing sites at once [8]. Most methods were largely ineffective and sometimes lowered a document’s ranking. Strategies that improved a source’s ranking in the model’s context did better. Gains also shrank as more sites adopted the same methods [8]. GEO is a real measurement. Its tactics are still being tested.

AEO: a label for the answer box

The phrase “answer engine” does appear in research. Narayanan Venkit and colleagues use it for LLM-based search that retrieves sources and writes a summary with citations, and they audited You.com, Perplexity and Bing Copilot [5]. In their automated evaluation of 303 queries, between about 23 and 32 percent of statements in each engine’s answers were not supported by the sources the engine listed [5]. An earlier audit of four generative search engines found that only about 51.5 percent of generated sentences were fully supported by their citations [6]. These studies are about answer engines. They are not studies of an optimization discipline called AEO. Because AEO and GEO point at the same surface, we use GEO, which has a definition and a benchmark behind it.

Why the three measures can disagree

On 14 May 2024, Google announced that AI Overviews would begin rolling out to everyone in the U.S. [4]. In that post, Google said links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query [4]. An independent Pew Research Center analysis looked at March 2025 browsing data from 900 U.S. adults [7]. Users who met an AI summary clicked a traditional result link in 8 percent of visits. Users who did not meet one clicked in 15 percent of visits. Clicks on links inside the summary happened in 1 percent of visits [7].

These two findings measure different things. Google compares an included link with the same page as a plain listing. Pew counts clicks per visit to a results page. Both can hold at once. That is the point of naming the measure before naming the tactic. In Pew’s sample, searches phrased as questions produced an AI summary more often: 60 percent of searches that began with words such as “who”, “what”, “when” or “why” did [7].

Map

Look at the map. One query can have two layers: a list of links and, sometimes, a written answer above it. In Pew’s sample, the answer layer appeared more often on longer, question-shaped searches [7].

Find your place. Read each query cluster three ways. Where does the page sit in the list (SEO)? Is it cited, and how much of the answer is drawn from it (GEO) [2]? Does the citation actually support the sentence it is attached to [5] [6]?

Align. Start with eligibility, because Google’s AI features require a page that is indexed and eligible for a snippet [3]. Then add what the GEO study measured: citations, quotations and relevant numbers [2]. Re-check over time, because both the engines and the evidence are moving [2] [8].

References

  1. Google Search Central. (n.d.). Search engine optimization (SEO) starter guide. Google for Developers. Accessed 9 October 2026. https://developers.google.com/search/docs/fundamentals/seo-starter-guide
  2. Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., & Deshpande, A. (2024). GEO: Generative engine optimization. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’24), 5–16. https://doi.org/10.1145/3637528.3671900 (preprint: https://arxiv.org/abs/2311.09735)
  3. Google Search Central. (n.d.). AI features and your website. Google for Developers. Accessed 9 October 2026. https://developers.google.com/search/docs/appearance/ai-features
  4. Reid, E. (2024, May 14). Generative AI in Search: Let Google do the searching for you. The Keyword, Google. https://blog.google/products/search/generative-ai-google-search-may-2024/
  5. Narayanan Venkit, P., Laban, P., Zhou, Y., Mao, Y., & Wu, C.-S. (2024). Search engines in an AI era: The false promise of factual and verifiable source-cited responses. arXiv:2410.22349. https://arxiv.org/abs/2410.22349
  6. Liu, N. F., Zhang, T., & Liang, P. (2023). Evaluating verifiability in generative search engines. Findings of the Association for Computational Linguistics: EMNLP 2023. https://aclanthology.org/2023.findings-emnlp.467/
  7. Chapekis, A., & Lieb, A. (2025, July 22). Google users are less likely to click on links when an AI summary appears in the results. Pew Research Center. https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
  8. Puerto, H., Gubri, M., Green, T., Oh, S. J., & Yun, S. (2025). C-SEO Bench: Does conversational SEO work? NeurIPS 2025 Datasets and Benchmarks Track. https://arxiv.org/abs/2506.11097

Which of the three readings do you want to stand in first? Tell us where you want to stand. Or continue with what an audit measures.