What Football Commentary Teaches Us About AEO: Why AI Engines Quote Some Sources and Ignore Others
Millions of World Cup questions are being answered by AI assistants this month — and the same few sources keep getting cited. The selection logic behind those citations is the clearest available guide to Answer Engine Optimisation.
By Robin Deane — Founder, RD
AI engines cite sources the way good football commentary quotes its pundits: they reach for whoever answers the exact question fastest, with authority, in liftable form. To become citable, structure pages so each section answers one question completely and quotably, establish verifiable expertise signals, keep facts current and dated, and stay technically accessible to AI crawlers. Being comprehensive is not the same as being quotable — and quotable is what gets cited.
Ask an AI assistant anything about the World Cup this month — who tops the group, how the new format works, what time kick-off is in your timezone — and you will get a confident answer with a handful of citations. Run those questions a few times and a pattern emerges: the same sources keep being quoted, while thousands of pages covering identical facts are never mentioned at all.
That pattern is the most public demonstration yet of how answer engines choose their sources — and understanding the selection logic is what Answer Engine Optimisation (AEO) actually is. The World Cup makes an unusually clean case study because the underlying information is identical everywhere: every outlet has the same scores, the same fixtures, the same rules. When the facts are commoditised, the citation decision is purely about how the source presents them. Which is exactly the situation most businesses are in with their competitors.
How Do AI Engines Actually Choose What to Cite?
Definition: Answer Engine Optimisation (AEO) is the practice of structuring content so AI-powered search experiences — ChatGPT search, Perplexity, Google AI Overviews, Copilot — select it as a source when composing answers. Where classic SEO competes for a ranked position a human clicks, AEO competes for being quoted, paraphrased, and linked inside a generated answer.
Watch football coverage closely and you will notice commentary teams do something structurally similar to answer engines. A moment happens; the commentator needs one sentence of authority — and reaches for the pundit who delivers exactly that: a crisp, complete, credentialed statement that slots straight into the broadcast. The rambling expert, however knowledgeable, does not get the airtime. The selection favours liftability, not just expertise.
Answer engines apply the same filter at scale. When one composes an answer, it retrieves candidate passages, then favours sources that are:
- Directly responsive — a passage that answers the specific question asked, not a page that covers the topic generally
- Self-contained — the passage makes sense lifted out of its surroundings, with the entities named rather than pronoun-referenced
- Authoritative on inspection — the source shows verifiable expertise signals: named authors, credentials, original data, consistent topical depth
- Current and dated — visibly fresh, with dates that let the engine judge whether the fact still holds
- Technically retrievable — accessible to AI crawlers, fast, and structured so extraction is unambiguous
Comprehensiveness — the traditional long-form SEO virtue — appears nowhere on that list. A 4,000-word page that buries every answer in flowing prose loses citations to a modest page that answers one question perfectly in its first eighty words.
What Do the Cited World Cup Sources Do Differently?
Compare the sources AI assistants keep citing for tournament questions against equally accurate sources they ignore, and the differences are consistently structural rather than substantive:
| Consistently cited | Consistently ignored |
|---|---|
| One question, one section — a heading that mirrors the question, answered completely in the first sentences beneath it | Everything woven together — answers exist but are distributed across narrative paragraphs |
| Facts stated with their context — "The 2026 final takes place on 19 July at MetLife Stadium" | Facts assuming context — "The final takes place there on the 19th" |
| Dated, maintained pages — visible update stamps on evolving information | Undated evergreen pages — the engine cannot tell if the fact survived the last round of matches |
| Structured data and clean markup — tables for fixtures, schema for events and FAQs | Visual-only structure — information locked in images, embeds, or scripts crawlers cannot parse |
| Named expertise — bylines, credentials, a track record on the topic | Anonymous aggregation — accurate, but indistinguishable from a thousand identical pages |
The lesson generalises with almost no translation: your industry's equivalent of match facts — pricing questions, how-tos, comparisons, definitions — is being asked to AI assistants right now, and the sources being quoted are the ones structured for lifting.
How Do You Make Your Content Citable?
Answer engines respond to conversational questions, which are longer and more specific than classic keywords. List the actual questions your buyers ask at every stage — then ask the engines themselves and note who gets cited today. That citation gap is your target list, and the research method is the same one behind a proper GEO/AEO programme.
Question-mirroring headings; a complete, self-contained answer in the first two or three sentences under each; detail and nuance after the answer, never instead of it. This piece — like every post on this site — follows exactly that pattern, from the Quick Answer at the top to the FAQ block below.
When facts are commoditised, original material wins citations: your own data, benchmarks, named-client results, genuine positions. Engines composing balanced answers actively seek distinctive, attributable perspectives — the pundit with a view gets quoted over the one reciting consensus.
Named authors with real footprints, About pages that establish who is behind the claims, schema markup (Article, FAQPage, Organization) that states structure explicitly, dates on everything that can go stale. E-E-A-T was always sound advice; answer engines made it mechanical.
Check robots.txt against the AI crawlers (GPTBot, ClaudeBot, PerplexityBot and peers) — some sites block them accidentally via blanket rules. Keep content server-rendered or statically generated; JavaScript-dependent content is still unevenly retrieved. An llms.txt file summarising the site helps engines orient. None of this is exotic: it is crawlability discipline with new user agents.
Does AEO Conflict With Classic SEO?
Mostly no — and where it does, the tension is worth naming honestly.
The overlap is large: crawlability, structured data, expertise signals, and genuinely useful content serve both. A page built as clean answer blocks tends to win featured snippets and rank respectably and get cited. The three real tensions:
Traffic versus presence. An AI answer that cites you may satisfy the user without a click. The visit is lost; the mention is not — brand presence inside answers builds the familiarity that later shows up as branded search and direct traffic, the same displaced-response pattern we described in second-screen measurement. Businesses selling considered purchases lose little (buyers of consequence click through); pure ad-impression publishers lose more.
Quotability versus narrative. Ruthlessly blocked answer-first pages can read mechanically. The resolution is layering, not choosing: answer first for the machines and the scanners, story after for the humans who stay. The best writers have always done this — journalism calls it the inverted pyramid.
Measurement maturity. Classic SEO has rank trackers; citation tracking is younger and noisier. It is measurable — periodic question-panels against the major engines, watching referral and branded-search lift — but expectations about precision should be set accordingly.
The strategic point is that the citation layer is being allocated now, while most competitors still optimise only for the blue links. Early sources that engines learn to trust get re-retrieved, re-cited, and reinforced — authority compounding in exactly the way early domain authority did two decades ago. During this World Cup, the pattern locked in within days: the same few sources, cited again and again, because they kept proving liftable. Your category's version of that race is running quietly right now. The commentary box has open chairs — but not for long, and they go to whoever gives the best sixty-second answer.
- AI engines select sources like commentators select pundits: for crisp, complete, credentialed, liftable statements — not for comprehensiveness
- The World Cup proves the point cleanly: identical facts everywhere, yet the same structurally superior sources win every citation
- Citable pages answer one question per section, completely, in the first sentences — with entities named and context self-contained
- Original data and genuine positions win when facts are commoditised; engines seek attributable perspectives
- Trust signals must be machine-verifiable: named authors, schema markup, visible dates, consistent topical depth
- Check that AI crawlers aren't blocked and content isn't JavaScript-locked — technical openness is table stakes
- Citation authority compounds like early-web domain authority; the allocation race in your category is happening now
Frequently Asked Questions
What is the difference between SEO, GEO, and AEO?
SEO (Search Engine Optimisation) targets ranked positions in classic search results that humans click. AEO (Answer Engine Optimisation) targets being selected and cited inside AI-generated answers — in ChatGPT search, Perplexity, Google AI Overviews, and Copilot. GEO (Generative Engine Optimisation) is often used interchangeably with AEO, or as the umbrella term for optimising visibility across all generative AI experiences. The disciplines share most fundamentals — crawlability, structure, expertise — but AEO specifically rewards passage-level quotability over page-level comprehensiveness.
How do you get cited by ChatGPT and Perplexity?
Structure content so individual passages answer specific questions completely and self-containedly; establish machine-verifiable expertise (named authors, schema, original data); keep pages dated and current; and ensure AI crawlers can access the site (no accidental robots.txt blocks, no JavaScript-locked content). Then verify empirically: ask the engines your target questions monthly, record who gets cited, and close the structural gaps between your pages and the winners. Citation follows retrievability plus quotability plus trust — all three, not any one.
Does being cited in AI answers actually drive business results?
Yes, through two routes. Direct: engines link their citations, and high-intent users click through to verify or go deeper — lower volume than classic search referrals but typically strong intent. Indirect and larger: repeated presence inside answers builds brand familiarity that surfaces later as branded search, direct visits, and being on the shortlist when the buyer moves — displaced response rather than last-click attribution. Businesses with considered purchases benefit most; measurement should watch branded-search and direct-traffic trends alongside referral data.
Should I block AI crawlers from my site?
For most businesses, no. Blocking GPTBot, ClaudeBot, PerplexityBot and peers removes you from the answer layer your buyers increasingly consult — conceding those citations to competitors. The case for blocking is limited to businesses whose entire model is monetising page visits that AI answers would substitute (some publishers), or protecting genuinely proprietary content. If in doubt, allow retrieval crawlers and revisit quarterly with data on how AI surfaces are referring to you.
How do you measure AEO performance?
Four practical layers: a monthly citation panel (a fixed list of target questions asked to each major engine, recording citations won and lost); referral traffic from AI surfaces where analytics can identify it; branded search and direct traffic trends as the displaced-response signal; and share-of-voice tooling as the category matures. Set expectations honestly — this is closer to early-2000s rank tracking than today's SEO dashboards — but directional measurement is entirely achievable and quickly reveals whether structural changes are winning citations.
Will AEO replace traditional SEO?
No — it is being added on top, and the two share most of their foundations. Classic ranked results still carry the majority of search behaviour, and the engines powering AI answers draw heavily on the same crawling and quality signals SEO has always cultivated. The practical posture: keep SEO fundamentals strong, restructure key pages as answer blocks (which helps both), add the AEO-specific layer — citation tracking, AI-crawler access, quotable original material — and treat every future content decision as serving both surfaces at once.
Keep Reading
