Frequently asked questions

It’s the set of editorial, semantic and technical practices that make a website correctly understood, correctly cited and correctly reused by large language models and the answer engines built on them (ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews). It isn’t a model hack — it’s editorial engineering applied to a new discovery channel.

SEO optimises for classic results pages ranked by blue links. AEO (Answer Engine Optimization) optimises for direct answers — featured snippets, voice assistants, answer panels. GEO (Generative Engine Optimization) optimises specifically for answers generated by AI engines. In practice all three share the same foundation (clear, structured, trustworthy, well-linked content), but GEO adds requirements specific to retrieval and generation.

It’s not a technical question, it’s a business question. Blocking them protects your content from AI training and crawling but excludes your brand from generated answers. Letting them through makes you visible but means your content is used without direct compensation. By default a visibility-driven publisher lets them through; a paywalled publication may decide differently. Our Technical page documents the trade-off in detail.

No. Classic SEO remains the foundation: without proper crawl, clean HTML and topical authority, no AI engine will cite you either. GEO doesn’t replace SEO — it extends it into new discovery channels with extra requirements around structure and citability.

Partially. Third-party tools (Profound, Peec AI, Otterly, Scrunch, AthenaHQ) run queries at regular intervals and measure the share of citations for a brand across a defined corpus. It’s a useful proxy, not an exhaustive metric. Citations vary with user, context, date and model version.

From a few weeks to several months, depending on your existing authority, how often each engine refreshes its index, and topical competitiveness. AI engines take longer than Google to pick up new content because their corpora are sometimes refreshed only quarterly.

Publishing llms.txt doesn’t hurt but isn’t a major lever today. No major engine uses it as a binding signal. Making it a priority one would misallocate effort. Priority stays on semantic structure, authority and crawlability.

Both, as long as it’s structured. What matters is passage legibility: a long article carved into standalone sections (clear H2/H3, 3–6 line paragraphs, standalone sentences) extracts as well as a short article. An unstructured long article is invisible despite its depth.

Indirectly. Schemas help engines identify entities, relationships and content types. This benefits classic SEO and Google AI Overviews first. LLMs with retrieval (ChatGPT, Perplexity) rely primarily on HTML and text: schema.org is not a substitute for editorial clarity.

Four recurring factors: 1) topical authority (the site is recognised on the subject), 2) freshness (content is recent or actively maintained), 3) extractability (the relevant passage can be lifted as-is), 4) sourcing (the content itself cites its sources, which models reward). None of these are hacks — they’re the outcome of serious editorial work.

Start with the audit checklist on this site: 40 points across the four dimensions of the LOOP framework (Legibility, Ontology, Operations, Performance). Complement with manual monitoring: run 10–20 strategic queries through ChatGPT, Perplexity, Gemini and Claude and look at who gets cited. You’ll have a useful baseline in half a day.

Yes. AI engines weight passage relevance above domain size. A small site with extremely clear content on a narrow niche is often cited ahead of a large site with vague content. Expertise and structure beat volume.

Primary: positions for 10 strategic queries in Google Search Console, impressions in the AI Overviews report, number of citations measured via a third-party tool. Secondary: topical backlinks, scroll depth on pillar pages, sign-ups against a deliverable (checklist, audit).

There’s both substance and hype. The substance: ChatGPT Search and AI Overviews capture a growing share of queries, and click-through to websites is declining on some intents. The hype: promises of “hacks” or “secrets” should be ignored. Investing in editorial quality and technical structure is a sound decision either way.

Not necessarily. Most of the work falls under extended SEO: audit, editorial plan, structure, technical. A dedicated consultant can be useful to set a method, train a team and bring benchmarks. They must not promise unmeasurable outcomes.

The foundation is the same, priorities differ. B2B: brand authority and comparison pages. E-commerce: structured product data (schema Product, Review), FAQs, help pages. Media: citability, freshness, article structure. Our Use Cases page details each profile.

1) Treating GEO as a subject separate from SEO rather than an extension. 2) Betting on llms.txt at the expense of the rest. 3) Padding pages with generic FAQs that add nothing. 4) Neglecting topical authority (links, mentions, co-occurrences). 5) Stacking schemas without handling validation errors. 6) Not measuring — or inventing metrics without a method.

Follow 3 to 5 primary sources (ChatGPT, Perplexity, Google blog, Anthropic, academic sources), not the noisy recaps. Refresh your pillar content every quarter. Keep an internal log of behavioural changes observed on your strategic queries.

An independent editorial project, not affiliated with any tool vendor or agency. The editorial line is publicly documented (positioning, method, sources). See the About page for details.

Yes, with source credit and a link to the page. Published frameworks (such as LOOP) are explicitly shareable under this condition. The site itself is designed to be cited by LLMs.