# AEO 101

Canonical: https://mudpie.ai/101/aeo/
Updated: 2026-09-20
Curator: Ali Abouelatta (https://mudpie.ai/authors/ali-abouelatta/)

What answer engines retrieve and cite, how unstable that evidence is, and which publishing changes deserve to be tested instead of repeated as folklore.

**TL;DR:** AEO is not one ranking system. Treat every result as engine-, prompt- and date-specific; preserve SEO fundamentals; and read vendor studies as hypotheses whose samples and denominators matter.

## Understand what is being measured

These sources separate generated-answer composition, citations, intent and user clicks—the metrics most AEO claims blur together.

### [GEO: Generative Engine Optimization](https://arxiv.org/abs/2311.09735)

Authority: [Pranjal Aggarwal](https://mudpie.ai/authors/pranjal-aggarwal/)

**TL;DR:** The original GEO paper tests how textual changes affect source visibility inside a fixed generative-engine benchmark; it does not prove a universal live-search ranking recipe.

**What you'll read:** The benchmark, visibility metric, nine editing methods, domain differences and the limits of a five-source GPT-3.5 simulation.

### [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features)

Authority: [Google Search Central](https://mudpie.ai/authors/google-search-central/)

**TL;DR:** Google says the same crawlability, indexability and people-first content foundations apply to AI Overviews and AI Mode; no special AI schema is required.

**What you'll read:** Official eligibility, preview controls, measurement guidance and the clearest line between supported optimization and invented AEO requirements.

### [AI platform citation patterns](https://www.tryprofound.com/blog/ai-platform-citation-patterns)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Different answer engines lean on different source stacks, so an aggregate 'AI visibility' score can hide the engine where a brand is actually absent.

**What you'll read:** A large cross-platform citation analysis, source-share comparisons and the methodological gaps that remain without the raw prompt sample.

### [Citation overlap strategy](https://www.tryprofound.com/blog/citation-overlap-strategy)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Low citation overlap between ChatGPT and Perplexity argues for engine-specific monitoring rather than one universal source strategy.

**What you'll read:** A 100,000-prompt domain-overlap study and the important distinction between shared domains, shared pages and shared answer claims.

### [AI Search Shift](https://www.tryprofound.com/blog/ai-search-shift)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** ChatGPT's source alignment with Google can change quickly, but even aligned citations do not behave like ordinary position-one search clicks.

**What you'll read:** Citation-alignment trends, a smaller position-overlap sample and a useful warning against translating ranking position directly into AI traffic.

## Respect volatility and behavior

Before acting on a visibility score, understand how much sources move and whether users click at all.

### [AI Search Volatility](https://www.tryprofound.com/blog/ai-search-volatility)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Large shares of cited domains changed across a one-month comparison, making one-run visibility snapshots unsafe planning inputs.

**What you'll read:** Repeated prompt samples across four platforms, domain-drift rates and the difference between citation churn and a brand disappearing entirely.

### [Introducing Prompt Research Reports](https://www.tryprofound.com/blog/introducing-prompt-research-reports-in-profound)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Prompt-volume claims depend on retrieval, clustering and coverage choices, so the product's own method description belongs beside any market-size number.

**What you'll read:** Profound's launch explanation of its conversation corpus, prompt selection and the moving denominators behind large reported totals.

### [Google users click less when an AI summary appears](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/)

Authority: [Pew Research Center](https://mudpie.ai/authors/pew-research-center/)

**TL;DR:** In Pew's browser-panel sample, search sessions with an AI summary produced fewer clicks to traditional results and very few clicks on cited links.

**What you'll read:** An independent behavioral sample, disclosed methods and a useful separation between summary exposure, citation presence and outbound clicking.

### [AIO impact on Google CTR: 2026 update](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update)

Authority: [Seer Interactive](https://mudpie.ai/authors/seer-interactive/)

**TL;DR:** Seer's longitudinal client data suggests AI summaries continue to reshape both paid and organic click-through rates, with branded demand behaving differently.

**What you'll read:** Observed CTR changes across query cohorts, the dataset's agency-client boundary and practical questions for search forecasting.

### [The data on Reddit and AI search](https://www.tryprofound.com/blog/the-data-on-reddit-and-ai-search)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Reddit is a meaningful but not dominant aggregate citation source; source type, topic and content age matter more than the slogan 'AI loves Reddit.'

**What you'll read:** A multi-billion-citation analysis, post-age findings, platform differences and the disclosure that Profound worked with Reddit.

## Turn evidence into a publishing plan

Use structure and prompt research as inputs to experiments—not guarantees. These readings show what to test and what still lacks causal proof.

### [How to optimize for answer engines](https://www.tryprofound.com/articles/how-to-optimize-answer-engines-2025)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Tables, headings, lists, schema and complete answers are plausible structural signals, but the public study does not publish a causal lift for any one change.

**What you'll read:** A 2,000-page pattern study that is useful as an experiment backlog when read with its missing controls and undisclosed query set.

### [ChatGPT intent landmark study](https://www.tryprofound.com/blog/chatgpt-intent-landmark-study)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** AI prompt intent does not map cleanly onto classic search intent, so keyword-derived prompt lists can misstate what people ask assistants to do.

**What you'll read:** A large vendor-classified prompt sample, its intent distribution and the missing details you need before treating it as a market census.

### [Prompt Research Reports](https://www.tryprofound.com/features/prompt-volumes/research-reports)

Authority: [Profound](https://mudpie.ai/authors/profound/)

**TL;DR:** Prompt research is a retrieval-and-clustering problem, not a direct equivalent of exact keyword volume from a search ad system.

**What you'll read:** Profound's own description of retrieval, ranking, clustering and coverage selection over its conversation corpus.

### [Beyond the SERP: Visibility Layer and Trust Stack](https://www.kevin-indig.com/talks/beyond-the-serp-visibility-trust)

Authority: [Kevin Indig](https://mudpie.ai/authors/kevin-indig/)

**TL;DR:** When citation clicks are rare, the strategic job expands from winning a click to becoming a trusted source across the surfaces that shape the answer.

**What you'll read:** A practical framework for zero-click visibility, source trust, brand mentions and measuring influence beyond sessions.

### [Winners and losers of the AI disruption of search](https://www.kevin-indig.com/talks/winners-losers-ai-search)

Authority: [Kevin Indig](https://mudpie.ai/authors/kevin-indig/)

**TL;DR:** AI search redistributes attention toward recognizable sources and away from interchangeable pages, making brand trust part of organic performance.

**What you'll read:** A synthesis of traffic shifts, trust signals, category differences and the strategic choices available to publishers and product companies.