What AEO actually is

Most agencies still describe AEO as a flavor of SEO. That framing is wrong, and it costs clients citations. Classical SEO is a list-of-links game. AEO is a quotation game. The unit of value in SEO is a ranked URL. The unit of value in AEO is a sentence the model decided to lift, attributed to your domain.

When a user asks ChatGPT a question, the model does not return ten blue links. It returns one composed answer. Sometimes there are citations in the margin, sometimes not. Either way, your business either appears inside that answer or it does not. AEO is the work of being the source the model reaches for.

Two numbers explain the urgency. ChatGPT crossed 800 million weekly active users in late 2025. Google AI Overviews now appear on a majority of informational queries in the United States, and clickthrough on the underlying organic results is down materially when an AI answer is present. Independent measurements put the CTR drop somewhere between 30 and 60 percent depending on query class.

If your business sells anything that starts with a research query, your funnel is being rewritten in real time. The companies being cited inside AI answers are catching the demand. The companies ranked beneath those answers are watching impressions stay flat while sessions fall.

There is a compounding window right now because AI engines reward consistent presence. Domains that get cited tend to get cited more. The flywheel is real, and it is being built right now by the firms that started in 2024 and 2025. Starting in 2027 means buying ground that is already taken.

The structural changes that win at AEO often also help SEO, but the writing model is different. SEO long-form pieces reward depth, internal linking, and topical authority. AEO rewards self-contained passages that can be quoted in isolation. A great SEO article is a 2,500-word survey of a topic. A great AEO article is the same survey with a 70-word answer block engineered into the top of every section.

The signals are also different. SEO leans on backlinks, on-page relevance, and Core Web Vitals. AEO leans on entity clarity, schema, passage structure, and crawler access for AI-specific user agents like GPTBot, PerplexityBot, ClaudeBot, and Google-Extended. Disallowing those bots in robots.txt removes you from the corpus entirely. Many sites still do this by accident.

The measurement model is different too. SEO tracks rank, impressions, clicks. AEO tracks citation count per platform per month, citation position within the answer, and conversion attribution from AI-referred traffic. The platforms do not give you Search Console for this, so the work involves real prompt panels, scrapers like the DataForSEO ChatGPT scraper, and disciplined monthly logging.

An FPWS AEO engagement starts with an audit. We pull the queries that matter, score the current AI answers across the major engines, identify who is being cited and on what page structure, and map the gaps. Then we rebuild the pages that need to win.

Passage citability is the first lever. Every H2 gets a 40 to 80 word citable answer block underneath it, written so that if the model lifts the block alone, the citation still makes sense. Blocks open with the entity, not a pronoun. They define the concept, then add one piece of context. They do not bury the answer two paragraphs deep.

The second lever is the technical layer. We ship an llms.txt at the root, an explicit robots.txt allow-list for the reputable AI crawlers, server-rendered HTML so models see the same content a user does, and JSON-LD schema with stable @id URIs across Organization, Person, Service, FAQPage, and Article. Schema is not a magic SEO sprinkle, but it is the structured spine that makes you a citable entity instead of a string of words.

The third lever is content production. We publish where there is a real query opportunity, not for volume. Each new piece is engineered to the same passage standard. Each gets the schema, the entity sameAs links, the bylined Person reference. Each is monitored after launch.

The fourth lever is monitoring. Monthly reports show citation count and position by engine, query, and competitor. We do not hand clients a vanity dashboard. We hand them a list: here are the citations you gained, here are the ones you lost, here is what we are publishing next month to extend the lead.

AEO is not keyword stuffing for ChatGPT. The models are large enough to detect manipulation and to rank sources for trust. AEO is not buying mentions on third-party sites in the hope they show up as citations, although strong third-party presence does help entity definition. AEO is not a one-time deliverable. The engines change their answer mix monthly, sometimes weekly, and the work has to keep up.

AEO is also not a replacement for SEO. Both disciplines compound. The same site benefits from ranking in classical search and being cited in AI answers, and the technical foundations overlap heavily. The wrong move is to pick one and ignore the other.

If your business sells anything researched online, the answer is now. First citations on a properly structured page typically appear within four to eight weeks. Compounding citation growth, the phase where the model reaches for you reflexively, takes three to six months of consistent publishing and monitoring. That is faster than classical SEO but not instant.

The cost of waiting is the cost of a competitor occupying the citation slot for the queries that drive your pipeline. Once a model has a preferred source for a question, displacing that source takes work. Better to be the preferred source.