What is answer engine optimization?

For twenty-five years, search worked one way: a query produced a list, and the job of optimization was to appear high on that list. The user did the synthesis themselves by clicking through options.

Answer engines removed that step. When someone asks ChatGPT which project management tool suits a small agency, they do not receive ten links. They receive a paragraph naming two or three, with the reasoning already done. Everything not named is not lower down the page — it is absent from the conversation entirely.

Research from Bain found that roughly 80% of searchers now rely on zero-click results for at least 40% of their searches, with associated organic traffic declining 15% to 25%. Pew Research measured the mechanism directly: when an AI summary appears, click-through on results below it falls from about 15% to 8%, and around 26% of those sessions end without any click at all.

The traffic did not move somewhere else. It stopped existing. What replaced it is a much smaller volume of far higher-intent referrals — Seer Interactive measured ChatGPT referral traffic converting at 15.9% against 1.76% for Google organic.

AEO work divides into four distinct layers, and confusing them is the most common reason engagements underperform.

The first is access. Answer engines can only cite content their crawlers can retrieve, which means GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended and others must be able to reach your site, and your content must exist in the initial HTML response rather than only after JavaScript executes.

The second is identity. These systems reason about entities rather than pages — organizations, people, products, and the relationships between them. A connected schema graph with consistent identifiers across every profile that mentions you lets a model resolve who you are with enough confidence to stake an answer on you.

The third is structure. Retrieval systems extract passages, not pages. Content shaped answer-first, under question-phrased headings, in chunks that remain coherent when pulled out of context, gets extracted; content that buries its answer in paragraph four does not.

The fourth is corroboration, and it is the one that decides outcomes. A model recommending a vendor is reproducing a consensus assembled from many sources. Building that consensus requires earned mentions in the places retrieval reaches — press, communities, review platforms, partner sites, video, directories.

This is a young enough discipline that a lot of advice circulating is inherited from SEO without being tested. Large-scale citation analysis suggests several inherited assumptions are wrong.

AEO does not replace SEO, and any agency claiming it does is selling a rebrand. Crawlability, indexation and structured data remain prerequisites — a page nothing can reach cannot be cited by anything.

What has changed is that the signals predicting AI citation are not the same as those predicting rankings, and a few point the opposite direction. That does not make links worthless; it means link volume is not the lever in this channel. The practical consequence is that AEO and SEO need one integrated strategy rather than two teams optimizing for divergent metrics.